docs: make Chinese README the default

.gitignore

@@ -10,6 +10,7 @@ gradlew
 gradlew.bat
 gradle
 local.properties
+gradle.properties
 
 # Android
 android/build

CMakeLists.txt

@@ -0,0 +1,21 @@
+cmake_minimum_required(VERSION 3.13)
+project(vosk-api CXX)
+
+set(CMAKE_CXX_STANDARD 17)
+set(CMAKE_CXX_EXTENSIONS OFF)
+
+add_library(vosk
+  src/language_model.cc
+  src/model.cc
+  src/recognizer.cc
+  src/spk_model.cc
+  src/vosk_api.cc
+  src/postprocessor.cc
+)
+
+find_package(kaldi REQUIRED)
+target_link_libraries(vosk PUBLIC kaldi-base kaldi-online2 kaldi-rnnlm fstngram)
+
+include(GNUInstallDirs)
+install(TARGETS vosk DESTINATION ${CMAKE_INSTALL_LIBDIR})
+install(FILES src/vosk_api.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})

README.en.md

@@ -0,0 +1,27 @@
+# Vosk Speech Recognition Toolkit
+
+Vosk is an offline open source speech recognition toolkit. It enables
+speech recognition for 20+ languages and dialects - English, Indian
+English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish,
+Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi, Filipino,
+Ukrainian, Kazakh, Swedish, Japanese, Esperanto, Hindi, Czech, Polish.
+More to come.
+
+Vosk models are small (50 Mb) but provide continuous large vocabulary
+transcription, zero-latency response with streaming API, reconfigurable
+vocabulary and speaker identification.
+
+Speech recognition bindings implemented for various programming languages
+like Python, Java, Node.JS, C#, C++, Rust, Go and others.
+
+Vosk supplies speech recognition for chatbots, smart home appliances,
+virtual assistants. It can also create subtitles for movies,
+transcription for lectures and interviews.
+
+Vosk scales from small devices like Raspberry Pi or Android smartphone to
+big clusters.
+
+# Documentation
+
+For installation instructions, examples and documentation visit [Vosk
+Website](https://alphacephei.com/vosk).

README.md

@@ -1,26 +1,21 @@
-# About
+<!-- WEHUB_ZH_README -->
+> [!NOTE]
+> 本文档由 WeHub 基于上游 README 翻译整理，属于社区翻译，非官方中文文档。  
+> [English](./README.en.md) · [原始项目](https://github.com/alphacep/vosk-api) · [上游 README](https://github.com/alphacep/vosk-api/blob/HEAD/README.md)  
+> 原作者、版权与许可证归属以原始项目及本仓库 LICENSE 文件为准。
 
-Vosk is an offline open source speech recognition toolkit. It enables
-speech recognition models for 18 languages and dialects - English, Indian
-English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish,
-Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi, Filipino,
-Ukrainian.
+# Vosk Speech Recognition Toolkit
 
-Vosk models are small (50 Mb) but provide continuous large vocabulary
-transcription, zero-latency response with streaming API, reconfigurable
-vocabulary and speaker identification.
+Vosk 是一款离线开源语音识别工具包。它支持 20 多种语言及方言的语音识别——英语、印度英语、德语、法语、西班牙语、葡萄牙语、中文、俄语、土耳其语、越南语、意大利语、荷兰语、加泰罗尼亚语、阿拉伯语、希腊语、波斯语、菲律宾语、乌克兰语、哈萨克语、瑞典语、日语、世界语、印地语、捷克语、波兰语。更多语言即将推出。
 
-Speech recognition bindings implemented for various programming languages
-like Python, Java, Node.JS, C#, C++ and others.
+Vosk 模型体积小（50 Mb），但可提供连续大词汇量转写、通过流式 API（streaming API）实现零延迟响应、可重新配置的词汇表以及说话人识别。
 
-Vosk supplies speech recognition for chatbots, smart home appliances,
-virtual assistants. It can also create subtitles for movies,
-transcription for lectures and interviews.
+已为 Python、Java、Node.JS、C#、C++、Rust、Go 等多种编程语言实现语音识别绑定。
 
-Vosk scales from small devices like Raspberry Pi or Android smartphone to
-big clusters.
+Vosk 为聊天机器人、智能家居设备、虚拟助手提供语音识别，还可为电影生成字幕，为讲座和访谈提供转写。
 
-# Documentation
+Vosk 的部署规模可从小型设备（如 Raspberry Pi 或 Android 智能手机）扩展到大型集群。
 
-For installation instructions, examples and documentation visit [Vosk
-Website](https://alphacephei.com/vosk).
+# 文档
+
+有关安装说明、示例和文档，请访问 [Vosk 网站](https://alphacephei.com/vosk).

README.wehub.md

@@ -0,0 +1,7 @@
+# WeHub 来源说明
+
+- 原始项目：`alphacep/vosk-api`
+- 原始仓库：https://github.com/alphacep/vosk-api
+- 导入方式：上游默认分支的最新快照
+- 原作者、版权和许可证信息以原始仓库及本仓库 LICENSE 为准
+- 本文件仅用于记录来源，不代表 WeHub 是原项目作者

android/build.gradle

@@ -1,57 +1,52 @@
 buildscript {
     repositories {
         google()
-        jcenter()
+        mavenCentral()
     }
     dependencies {
-        classpath 'com.android.tools.build:gradle:4.1.3'
+        classpath 'com.android.tools.build:gradle:8.13.0'
+        classpath 'com.vanniktech:gradle-maven-publish-plugin:0.34.0'
     }
 }
 
 allprojects {
-    version = '0.3.30'
+    version = '0.3.75'
 }
 
 subprojects {
 
     apply plugin: 'com.android.library'
-    apply plugin: 'maven-publish'
+    apply plugin: 'com.vanniktech.maven.publish'
 
     repositories {
         google()
-        jcenter()
+        mavenCentral()
     }
 
-    publishing {
-        publications {
-            aar(MavenPublication) {
-                groupId 'com.alphacephei'
-                version version
-                pom {
-                    url = 'http://www.alphacephei.com.com/vosk/'
-                        licenses {
-                            license {
-                            name = 'The Apache License, Version 2.0'
-                            url = 'http://www.apache.org/licenses/LICENSE-2.0.txt'
-                        }
-                    }
-                    developers {
-                        developer {
-                            id = 'com.alphacephei'
-                            name = 'Alpha Cephei Inc'
-                            email = 'contact@alphacephei.com'
-                        }
-                    }
-                    scm {
-                        connection = 'scm:git:git://github.com/alphacep/vosk-api.git'
-                        url = 'https://github.com/alphacep/vosk-api/'
-                    }
+    mavenPublishing {
+        publishToMavenCentral()
+        signAllPublications()
+    }
+
+    mavenPublishing {
+        pom {
+            url = 'http://www.alphacephei.com.com/vosk/'
+            licenses {
+                license {
+                    name = 'The Apache License, Version 2.0'
+                    url = 'http://www.apache.org/licenses/LICENSE-2.0.txt'
                 }
             }
-        }
-        repositories {
-            maven {
-                url = "$rootDir/repo"
+            developers {
+                developer {
+                    id = 'com.alphacephei'
+                    name = 'Alpha Cephei Inc'
+                    email = 'contact@alphacephei.com'
+                }
+            }
+            scm {
+                connection = 'scm:git:git://github.com/alphacep/vosk-api.git'
+                url = 'https://github.com/alphacep/vosk-api/'
             }
         }
     }

android/lib/build-vosk.sh

@@ -29,7 +29,7 @@ set -x
 OS_NAME=`echo $(uname -s) | tr '[:upper:]' '[:lower:]'`
 ANDROID_TOOLCHAIN_PATH=$ANDROID_NDK_HOME/toolchains/llvm/prebuilt/${OS_NAME}-x86_64
 WORKDIR_BASE=`pwd`/build
-PATH=$PATH:$ANDROID_NDK_HOME/toolchains/llvm/prebuilt/${OS_NAME}-x86_64/bin
+PATH=$ANDROID_TOOLCHAIN_PATH/bin:$PATH
 OPENFST_VERSION=1.8.0
 
 for arch in armeabi-v7a arm64-v8a x86_64 x86; do
@@ -40,34 +40,43 @@ case $arch in
     armeabi-v7a)
           BLAS_ARCH=ARMV7
           HOST=arm-linux-androideabi
-          AR=arm-linux-androideabi-ar
+          AR=llvm-ar
+          RANLIB=llvm-ranlib
           CC=armv7a-linux-androideabi21-clang
           CXX=armv7a-linux-androideabi21-clang++
           ARCHFLAGS="-mfloat-abi=softfp -mfpu=neon"
+          PAGESIZE_LDFLAGS=""
           ;;
     arm64-v8a)
           BLAS_ARCH=ARMV8
           HOST=aarch64-linux-android
-          AR=aarch64-linux-android-ar
+          AR=llvm-ar
+          RANLIB=llvm-ranlib
           CC=aarch64-linux-android21-clang
           CXX=aarch64-linux-android21-clang++
           ARCHFLAGS=""
+          # Ensure compatibility with 16KiB page size devices
+          PAGESIZE_LDFLAGS="-Wl,-z,common-page-size=4096 -Wl,-z,max-page-size=16384"
           ;;
     x86_64)
           BLAS_ARCH=ATOM
           HOST=x86_64-linux-android
-          AR=x86_64-linux-android-ar
+          AR=llvm-ar
+          RANLIB=llvm-ranlib
           CC=x86_64-linux-android21-clang
           CXX=x86_64-linux-android21-clang++
           ARCHFLAGS=""
+          PAGESIZE_LDFLAGS=""
           ;;
     x86)
           BLAS_ARCH=ATOM
           HOST=i686-linux-android
-          AR=i686-linux-android-ar
+          AR=llvm-ar
+          RANLIB=llvm-ranlib
           CC=i686-linux-android21-clang
           CXX=i686-linux-android21-clang++
           ARCHFLAGS=""
+          PAGESIZE_LDFLAGS=""
           ;;
 esac
 
@@ -75,16 +84,16 @@ mkdir -p $WORKDIR/local/lib
 
 # openblas first
 cd $WORKDIR
-git clone -b v0.3.13 --single-branch https://github.com/xianyi/OpenBLAS
-make -C OpenBLAS TARGET=$BLAS_ARCH ONLY_CBLAS=1 AR=$AR CC=$CC HOSTCC=gcc ARM_SOFTFP_ABI=1 USE_THREAD=0 NUM_THREADS=1 -j4
+git clone -b v0.3.20 --single-branch https://github.com/xianyi/OpenBLAS
+make -C OpenBLAS TARGET=$BLAS_ARCH ONLY_CBLAS=1 AR=$AR CC=$CC HOSTCC=gcc ARM_SOFTFP_ABI=1 USE_THREAD=0 NUM_THREADS=1 -j 8
 make -C OpenBLAS install PREFIX=$WORKDIR/local
 
 # CLAPACK
 cd $WORKDIR
 git clone -b v3.2.1  --single-branch https://github.com/alphacep/clapack
 mkdir -p clapack/BUILD && cd clapack/BUILD
-cmake -DCMAKE_C_FLAGS=$ARCHFLAGS -DCMAKE_C_COMPILER_TARGET=$HOST \
-    -DCMAKE_C_COMPILER=$CC -DCMAKE_SYSTEM_NAME=Generic -DCMAKE_AR=$ANDROID_NDK_HOME/toolchains/llvm/prebuilt/${OS_NAME}-x86_64/bin/$AR \
+cmake -DCMAKE_C_FLAGS="$ARCHFLAGS" -DCMAKE_C_COMPILER_TARGET=$HOST \
+    -DCMAKE_C_COMPILER=$CC -DCMAKE_SYSTEM_NAME=Generic -DCMAKE_AR=$ANDROID_TOOLCHAIN_PATH/bin/$AR \
     -DCMAKE_TRY_COMPILE_TARGET_TYPE=STATIC_LIBRARY \
     -DCMAKE_CROSSCOMPILING=True ..
 make -j 8 -C F2CLIBS/libf2c
@@ -105,25 +114,27 @@ make install
 
 # Kaldi itself
 cd $WORKDIR
-git clone -b android-mix --single-branch https://github.com/alphacep/kaldi
+git clone -b vosk-android --single-branch https://github.com/alphacep/kaldi
 cd $WORKDIR/kaldi/src
-CXX=$CXX CXXFLAGS="$ARCHFLAGS -O3 -DFST_NO_DYNAMIC_LINKING" ./configure --use-cuda=no \
+CXX=$CXX AR=$AR RANLIB=$RANLIB CXXFLAGS="$ARCHFLAGS -O3 -DFST_NO_DYNAMIC_LINKING" ./configure --use-cuda=no \
     --mathlib=OPENBLAS_CLAPACK --shared \
     --android-incdir=${ANDROID_TOOLCHAIN_PATH}/sysroot/usr/include \
     --host=$HOST --openblas-root=${WORKDIR}/local \
     --fst-root=${WORKDIR}/local --fst-version=${OPENFST_VERSION}
 make -j 8 depend
 cd $WORKDIR/kaldi/src
-make -j 8 online2 lm rnnlm
+make -j 8 online2 rnnlm
 
 # Vosk-api
 cd $WORKDIR
-#rm -rf vosk-api
-git clone -b master --single-branch https://github.com/alphacep/vosk-api
-cd vosk-api/src
-make -j 8 KALDI_ROOT=${WORKDIR}/kaldi OPENFST_ROOT=${WORKDIR}/local OPENBLAS_ROOT=${WORKDIR}/local CXX=$CXX EXTRA_LDFLAGS="-llog -static-libstdc++"
-
-# Copy JNI library to sources
-cp $WORKDIR/vosk-api/src/libvosk.so $WORKDIR/../../src/main/jniLibs/$arch/libvosk.so
+mkdir -p $WORKDIR/vosk
+make -j 8 -C ${WORKDIR_BASE}/../../../src \
+    OUTDIR=$WORKDIR/vosk \
+    KALDI_ROOT=${WORKDIR}/kaldi \
+    OPENFST_ROOT=${WORKDIR}/local \
+    OPENBLAS_ROOT=${WORKDIR}/local \
+    CXX=$CXX \
+    EXTRA_LDFLAGS="-llog -static-libstdc++ -Wl,-soname,libvosk.so ${PAGESIZE_LDFLAGS}"
+cp $WORKDIR/vosk/libvosk.so $WORKDIR/../../src/main/jniLibs/$arch/libvosk.so
 
 done

android/lib/build.gradle

@@ -3,13 +3,15 @@ def pomName = "Vosk Android"
 def pomDescription = "Vosk speech recognition library for Android"
 
 android {
-    compileSdkVersion 29
+    namespace 'org.vosk'
+    compileSdkVersion 36
     defaultConfig {
         minSdkVersion 21
-        targetSdkVersion 29
-        versionCode 6
+        targetSdkVersion 36
+        versionCode 10
         versionName = version
         archivesBaseName = archiveName
+        ndkVersion = "28.2.13676358"
     }
     compileOptions {
         sourceCompatibility JavaVersion.VERSION_1_8
@@ -19,24 +21,19 @@ android {
 
 task buildVosk(type: Exec) {
     commandLine './build-vosk.sh'
-    environment ANDROID_NDK_HOME: android.getSdkDirectory()
+    environment ANDROID_NDK_HOME: android.getNdkDirectory()
 }
 
 dependencies {
-    implementation 'net.java.dev.jna:jna:4.4.0@aar'
+    api 'net.java.dev.jna:jna:5.18.1@aar'
 }
 
 //preBuild.dependsOn buildVosk
 
-publishing {
-    publications {
-        aar(MavenPublication) {
-            artifactId = archiveName
-            artifact("$buildDir/outputs/aar/$archiveName-release.aar")
-            pom {
-                name = pomName
-                description = pomDescription
-            }
-        }
+mavenPublishing {
+    coordinates("com.alphacephei", archiveName, version)
+    pom {
+        name = pomName
+        description = pomDescription
     }
 }

android/lib/settings.gradle

@@ -1 +0,0 @@
-include 'model-en'

android/lib/src/main/AndroidManifest.xml

@@ -1,3 +1,3 @@
 <?xml version="1.0" encoding="utf-8"?>
-<manifest xmlns:android="http://schemas.android.com/apk/res/android" package="org.vosk">
+<manifest>
 </manifest>

android/lib/src/main/java/org/vosk/LibVosk.java

@@ -36,6 +36,8 @@ public class LibVosk {
 
     public static native void vosk_recognizer_set_words(Pointer recognizer, boolean words);
 
+    public static native void vosk_recognizer_set_partial_words(Pointer recognizer, boolean partial_words);
+
     public static native void vosk_recognizer_set_spk_model(Pointer recognizer, Pointer spk_model);
 
     public static native boolean vosk_recognizer_accept_waveform(Pointer recognizer, byte[] data, int len);
@@ -50,10 +52,30 @@ public class LibVosk {
 
     public static native String vosk_recognizer_partial_result(Pointer recognizer);
 
+    public static native void vosk_recognizer_set_grm(Pointer recognizer, String grammar);
+
     public static native void vosk_recognizer_reset(Pointer recognizer);
 
+    public static native void vosk_recognizer_set_endpointer_mode(Pointer recognizer, int mode);
+
+    public static native void vosk_recognizer_set_endpointer_delays(Pointer recognizer, float t_start_max, float t_end, float t_max);
+
     public static native void vosk_recognizer_free(Pointer recognizer);
 
+    public static native Pointer vosk_text_processor_new(String verbalizer, String tagger);
+
+    public static native void vosk_text_processor_free(Pointer processor);
+
+    public static native String vosk_text_processor_itn(Pointer processor, String input);
+
+    /**
+     * Set log level for Kaldi messages.
+     *
+     *  @param loglevel the level
+     *     0 - default value to print info and error messages but no debug
+     *     less than 0 - don't print info messages
+     *     greater than 0 - more verbose mode
+     */
     public static void setLogLevel(LogLevel loglevel) {
         vosk_set_log_level(loglevel.getValue());
     }

android/lib/src/main/java/org/vosk/Model.java

@@ -1,13 +1,18 @@
 package org.vosk;
 
+import java.io.IOException;
 import com.sun.jna.PointerType;
 
 public class Model extends PointerType implements AutoCloseable {
     public Model() {
     }
 
-    public Model(String path) {
+    public Model(String path) throws IOException {
         super(LibVosk.vosk_model_new(path));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a model");
+        }
     }
 
     @Override

android/lib/src/main/java/org/vosk/Recognizer.java

@@ -1,32 +1,163 @@
 package org.vosk;
 
 import com.sun.jna.PointerType;
+import java.io.IOException;
 
 public class Recognizer extends PointerType implements AutoCloseable {
-    public Recognizer(Model model, float sampleRate) {
+    /**
+     * Creates the recognizer object.
+     *
+     * The recognizers process the speech and return text using shared model data
+     * @param model       VoskModel containing static data for recognizer. Model can be
+     *                    shared across recognizers, even running in different threads.
+     * @param sampleRate The sample rate of the audio you are going to feed into the recognizer.
+     *                    Make sure this rate matches the audio content, it is a common
+     *                    issue causing accuracy problems.
+     * @throws IOException if the recognizer could not be created
+     */
+    public Recognizer(Model model, float sampleRate) throws IOException {
         super(LibVosk.vosk_recognizer_new(model, sampleRate));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a recognizer");
+        }
     }
 
-    public Recognizer(Model model, float sampleRate, SpeakerModel spkModel) {
+    /**
+     * Creates the recognizer object with speaker recognition.
+     *
+     *  With the speaker recognition mode the recognizer not just recognize
+     *  text but also return speaker vectors one can use for speaker identification
+     *
+     *  @param model       VoskModel containing static data for recognizer. Model can be
+     *                     shared across recognizers, even running in different threads.
+     *  @param sampleRate The sample rate of the audio you are going to feed into the recognizer.
+     *                     Make sure this rate matches the audio content, it is a common
+     *                     issue causing accuracy problems.
+     *  @param spkModel speaker model for speaker identification
+     *  @throws IOException if the recognizer could not be created
+     */
+    public Recognizer(Model model, float sampleRate, SpeakerModel spkModel) throws IOException {
         super(LibVosk.vosk_recognizer_new_spk(model.getPointer(), sampleRate, spkModel.getPointer()));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a recognizer");
+        }
     }
 
-    public Recognizer(Model model, float sampleRate, String grammar) {
+    /**
+     * Creates the recognizer object with the phrase list.
+     *
+     *  Sometimes when you want to improve recognition accuracy and when you don't need
+     *  to recognize large vocabulary you can specify a list of phrases to recognize. This
+     *  will improve recognizer speed and accuracy but might return [unk] if user said
+     *  something different.
+     *
+     *  Only recognizers with lookahead models support this type of quick configuration.
+     *  Precompiled HCLG graph models are not supported.
+     *
+     *  @param model       VoskModel containing static data for recognizer. Model can be
+     *                     shared across recognizers, even running in different threads.
+     *  @param sampleRate The sample rate of the audio you are going to feed into the recognizer.
+     *                     Make sure this rate matches the audio content, it is a common
+     *                     issue causing accuracy problems.
+     *  @param grammar The string with the list of phrases to recognize as JSON array of strings,
+     *                 for example "["one two three four five", "[unk]"]".
+     *  @throws IOException if the recognizer could not be created
+     */
+    public Recognizer(Model model, float sampleRate, String grammar) throws IOException {
         super(LibVosk.vosk_recognizer_new_grm(model.getPointer(), sampleRate, grammar));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a recognizer");
+        }
     }
 
+    /**
+     * Configures recognizer to output n-best results.
+     *
+     * <pre>
+     *   {
+     *      "alternatives": [
+     *          { "text": "one two three four five", "confidence": 0.97 },
+     *          { "text": "one two three for five", "confidence": 0.03 },
+     *      ]
+     *   }
+     * </pre>
+     *
+     * @param maxAlternatives - maximum alternatives to return from recognition results
+     */
     public void setMaxAlternatives(int maxAlternatives) {
         LibVosk.vosk_recognizer_set_max_alternatives(this.getPointer(), maxAlternatives);
     }
 
+    /** Enables words with times in the output
+     *
+     * <pre>
+     *   "result" : [{
+     *       "conf" : 1.000000,
+     *       "end" : 1.110000,
+     *       "start" : 0.870000,
+     *       "word" : "what"
+     *     }, {
+     *       "conf" : 1.000000,
+     *       "end" : 1.530000,
+     *       "start" : 1.110000,
+     *       "word" : "zero"
+     *     }, {
+     *       "conf" : 1.000000,
+     *       "end" : 1.950000,
+     *       "start" : 1.530000,
+     *       "word" : "zero"
+     *     }, {
+     *       "conf" : 1.000000,
+     *       "end" : 2.340000,
+     *       "start" : 1.950000,
+     *       "word" : "zero"
+     *     }, {
+     *       "conf" : 1.000000,
+     *       "end" : 2.610000,
+     *       "start" : 2.340000,
+     *       "word" : "one"
+     *     }],
+     * </pre>
+     *
+     * @param words - boolean value
+     */
     public void setWords(boolean words) {
         LibVosk.vosk_recognizer_set_words(this.getPointer(), words);
     }
 
+    /**
+     * Like above return words and confidences in partial results.
+     *
+     * @param partial_words - boolean value
+     */
+    public void setPartialWords(boolean partial_words) {
+        LibVosk.vosk_recognizer_set_partial_words(this.getPointer(), partial_words);
+    }
+
+    /**
+     * Adds speaker model to already initialized recognizer.
+     *
+     * Can add speaker recognition model to already created recognizer.
+     * Helps to initialize speaker recognition for grammar-based recognizer.
+     *
+     * @param spkModel Speaker recognition model
+     */
     public void setSpeakerModel(SpeakerModel spkModel) {
         LibVosk.vosk_recognizer_set_spk_model(this.getPointer(), spkModel.getPointer());
     }
 
+    /**
+     * Accept and process new chunk of voice data.
+     *
+     *  @param data - audio data in PCM 16-bit mono format
+     *  @param len - length of the audio data
+     *  @return 1 if silence is occurred and you can retrieve a new utterance with result method
+     *           0 if decoding continues
+     *           -1 if exception occurred
+     */
     public boolean acceptWaveForm(byte[] data, int len) {
         return LibVosk.vosk_recognizer_accept_waveform(this.getPointer(), data, len);
     }
@@ -39,22 +170,104 @@ public class Recognizer extends PointerType implements AutoCloseable {
         return LibVosk.vosk_recognizer_accept_waveform_f(this.getPointer(), data, len);
     }
 
+    /**
+     * Returns speech recognition result
+     *
+     * @return the result in JSON format which contains decoded line, decoded
+     *          words, times in seconds and confidences. You can parse this result
+     *          with any json parser
+     *
+     * <pre>
+     *  {
+     *    "text" : "what zero zero zero one"
+     *  }
+     * </pre>
+     *
+     * If alternatives enabled it returns result with alternatives, see also #setMaxAlternatives().
+     *
+     * If word times enabled returns word time, see also #setWordTimes().
+     */
     public String getResult() {
         return LibVosk.vosk_recognizer_result(this.getPointer());
     }
 
+    /**
+     * Returns partial speech recognition.
+     *
+     * @return partial speech recognition text which is not yet finalized.
+     *          result may change as recognizer process more data.
+     *
+     * <pre>
+     * {
+     *    "partial" : "cyril one eight zero"
+     * }
+     * </pre>
+     */
     public String getPartialResult() {
         return LibVosk.vosk_recognizer_partial_result(this.getPointer());
     }
 
+    /**
+     * Returns speech recognition result. Same as result, but doesn't wait for silence.
+     *  You usually call it in the end of the stream to get final bits of audio. It
+     *  flushes the feature pipeline, so all remaining audio chunks got processed.
+     *
+     *  @return speech result in JSON format.
+     */
     public String getFinalResult() {
         return LibVosk.vosk_recognizer_final_result(this.getPointer());
     }
 
+    /**
+     * Reconfigures recognizer to use grammar.
+     *
+     * @param grammar      Set of phrases in JSON array of strings or "[]" to use default model graph.
+     * @see #Recognizer(Model, float, String)
+     */
+    public void setGrammar(String grammar) {
+        LibVosk.vosk_recognizer_set_grm(this.getPointer(), grammar);
+    }
+
+    /**
+     * Resets the recognizer.
+     * Resets current results so the recognition can continue from scratch.
+     */
     public void reset() {
         LibVosk.vosk_recognizer_reset(this.getPointer());
     }
 
+    /**
+     * Endpointer delay mode
+     */
+    public class EndpointerMode {
+        public static final int DEFAULT = 0;
+        public static final int SHORT = 1;
+        public static final int LONG = 2;
+        public static final int VERY_LONG = 3;
+    }
+
+    /**
+     * Configures endpointer mode for recognizer
+     */
+    public void setEndpointerMode(int mode) {
+        LibVosk.vosk_recognizer_set_endpointer_mode(this.getPointer(), mode);
+    }
+
+    /**
+     * Set endpointer delays
+     *
+     * @param t_start_max     timeout for stopping recognition in case of initial silence (usually around 5.0)
+     * @param t_end           timeout for stopping recognition in milliseconds after we recognized something (usually around 0.5 - 1.0)
+     * @param t_max           timeout for forcing utterance end in milliseconds (usually around 20-30)
+     **/
+    public void setEndpointerDelays(float t_start_max, float t_end, float t_max) {
+        LibVosk.vosk_recognizer_set_endpointer_delays(this.getPointer(), t_start_max, t_end, t_max);
+    }
+
+    /**
+     * Releases recognizer object.
+     * Underlying model is also unreferenced and if needed, released.
+     */
     @Override
     public void close() {
         LibVosk.vosk_recognizer_free(this.getPointer());

android/lib/src/main/java/org/vosk/SpeakerModel.java

@@ -1,13 +1,36 @@
 package org.vosk;
 
 import com.sun.jna.PointerType;
+import java.io.IOException;
 
+/**
+ * Helps to initialize speaker recognition for grammar-based recognizer.
+ */
 public class SpeakerModel extends PointerType implements AutoCloseable {
     public SpeakerModel() {
     }
 
-    public SpeakerModel(String path) {
+    /**
+     * Loads speaker model data from the file.
+     *
+     * The path must contain:
+     * - a config file: mfcc.conf
+     * - kaldi nnet: final.ext.raw
+     * - mean.vec
+     * - transform.mat
+     *
+     * @param path the path of the model on the filesystem
+     * @throws IOException if the model could not be created
+     *
+     * @see <a href="http://kaldi-asr.org/doc/structkaldi_1_1MfccOptions.html">Kaldi MfccOptions</a>
+     * @see <a href="http://kaldi-asr.org/doc/classkaldi_1_1nnet3_1_1Nnet.html">Kaldi Nnet</a>
+     */
+    public SpeakerModel(String path) throws IOException {
         super(LibVosk.vosk_spk_model_new(path));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a speaker model");
+        }
     }
 
     @Override

android/lib/src/main/java/org/vosk/TextProcessor.java

@@ -0,0 +1,21 @@
+package org.vosk;
+
+import com.sun.jna.PointerType;
+
+public class TextProcessor extends PointerType implements AutoCloseable {
+    public TextProcessor() {
+    }
+
+    public TextProcessor(String verbalizer, String tagger) {
+        super(LibVosk.vosk_text_processor_new(verbalizer, tagger));
+    }
+
+    @Override
+    public void close() {
+        LibVosk.vosk_text_processor_free(this.getPointer());
+    }
+
+    public String itn(String input) {
+        return LibVosk.vosk_text_processor_itn(this.getPointer(), input);
+    }
+}

android/lib/src/main/java/org/vosk/android/SpeechService.java

@@ -19,9 +19,9 @@ import android.media.AudioRecord;
 import android.media.MediaRecorder.AudioSource;
 import android.os.Handler;
 import android.os.Looper;
+import android.annotation.SuppressLint;
 
 import org.vosk.Recognizer;
-
 import java.io.IOException;
 
 /**
@@ -48,6 +48,7 @@ public class SpeechService {
      *
      * @throws IOException thrown if audio recorder can not be created for some reason.
      */
+    @SuppressLint("MissingPermission")
     public SpeechService(Recognizer recognizer, float sampleRate) throws IOException {
         this.recognizer = recognizer;
         this.sampleRate = (int) sampleRate;
@@ -65,6 +66,51 @@ public class SpeechService {
         }
     }
 
+    /**
+     * Creates speech service with a caller-supplied {@link AudioRecord}.
+     * <p>
+     * Use this when you need to control the audio input device - for example,
+     * to pin recording to the built-in microphone when an external USB device
+     * without a microphone is present:
+     * <pre>
+     * AudioRecord recorder = new AudioRecord.Builder()
+     *         .setAudioSource(MediaRecorder.AudioSource.VOICE_RECOGNITION)
+     *         .setAudioFormat(format)
+     *         .build();
+     * if (Build.VERSION.SDK_INT >= 28) {
+     *     AudioManager am = (AudioManager) context.getSystemService(Context.AUDIO_SERVICE);
+     *     for (AudioDeviceInfo d : am.getDevices(AudioManager.GET_DEVICES_INPUTS)) {
+     *         if (d.getType() == AudioDeviceInfo.TYPE_BUILTIN_MIC) {
+     *             recorder.setPreferredDevice(d);
+     *             break;
+     *         }
+     *     }
+     * }
+     * SpeechService service = new SpeechService(recognizer, 16000f, recorder);
+     * </pre>
+     * <p>
+     * The caller retains ownership of {@code recorder}: if this constructor
+     * throws, the recorder is <em>not</em> released. Call
+     * {@link AudioRecord#release()} yourself in that case.
+     *
+     * @param recognizer the Vosk recognizer
+     * @param sampleRate sample rate in Hz; must match {@code recorder}'s configuration
+     * @param recorder   a fully-initialised {@link AudioRecord}
+     * @throws IOException if {@code recorder} is in STATE_UNINITIALIZED
+     */
+    public SpeechService(Recognizer recognizer, float sampleRate, AudioRecord recorder)
+            throws IOException {
+        this.recognizer = recognizer;
+        this.sampleRate = (int) sampleRate;
+        this.recorder   = recorder;
+
+        bufferSize = Math.round(this.sampleRate * BUFFER_SIZE_SECONDS);
+
+        if (recorder.getState() == AudioRecord.STATE_UNINITIALIZED) {
+            throw new IOException(
+                    "Failed to initialize recorder. Microphone might be already in use.");
+        }
+    }
 
     /**
      * Starts recognition. Does nothing if recognition is active.
@@ -136,6 +182,19 @@ public class SpeechService {
         return stopRecognizerThread();
     }
 
+
+    /**
+     * Returns the audio session ID of the underlying {@link AudioRecord}.
+     * <p>
+     * The session ID can be used to attach audio effects such as
+     * {@link android.media.audiofx.NoiseSuppressor} to the recording session.
+     *
+     * @return audio session ID, or {@link AudioRecord#ERROR} if unavailable
+     */
+    public int getAudioSessionId() {
+        return recorder.getAudioSessionId();
+    }
+
     /**
      * Shutdown the recognizer and release the recorder
      */

android/model-en/build.gradle

@@ -3,11 +3,12 @@ def pomName = "Vosk English Model"
 def pomDescription = "Small English model for Android"
 
 android {
-    compileSdkVersion 29
+    namespace "org.vosk"
+    compileSdkVersion 36
     defaultConfig {
         minSdkVersion 21
-        targetSdkVersion 29
-        versionCode 6
+        targetSdkVersion 36
+        versionCode 10
         versionName = version
         archivesBaseName = archiveName
     }
@@ -33,15 +34,10 @@ tasks.register('genUUID') {
 
 preBuild.dependsOn(genUUID)
 
-publishing {
-    publications {
-        aar(MavenPublication) {
-            artifactId = archiveName
-            artifact("$buildDir/outputs/aar/$archiveName-release.aar")
-            pom {
-                name = pomName
-                description = pomDescription
-            }
-        }
+mavenPublishing {
+    coordinates("com.alphacephei", archiveName, version)
+    pom {
+        name = pomName
+        description = pomDescription
     }
 }

android/model-en/src/main/AndroidManifest.xml

@@ -1,3 +1,2 @@
-<manifest xmlns:android="http://schemas.android.com/apk/res/android"
-    package="org.vosk.model.en">
+<manifest>
 </manifest>

android/model-en/src/main/assets/model-en-us/README

@@ -1,7 +0,0 @@
-US English model for mobile Vosk applications
-
-Copyright 2020 Alpha Cephei Inc
-
-Accuracy: 10.38 (tedlium test) 9.85 (librispeech test-clean)
-Speed: 0.11xRT (desktop)
-Latency: 0.15s (right context)

c/Makefile

@@ -1,16 +1,16 @@
 CFLAGS=-I../src
-LDFLAGS=-L../src -lvosk -ldl -lpthread -Wl,-rpath=../src
+LDFLAGS=-L../src -lvosk -ldl -lpthread -Wl,-rpath,../src
 
 all: test_vosk test_vosk_speaker
 
 test_vosk: test_vosk.o
-	g++ $^ -o $@ $(LDFLAGS)
+	gcc $^ -o $@ $(LDFLAGS)
 
 test_vosk_speaker: test_vosk_speaker.o
-	g++ $^ -o $@ $(LDFLAGS)
+	gcc $^ -o $@ $(LDFLAGS)
 
 %.o: %.c
-	g++ $(CFLAGS) -c -o $@ $<
+	gcc $(CFLAGS) -c -o $@ $<
 
 clean:
 	rm -f *.o *.a test_vosk test_vosk_speaker

csharp/demo/VoskDemo.cs

@@ -28,13 +28,16 @@ public class VoskDemo
    {
         // Demo float array
         VoskRecognizer rec = new VoskRecognizer(model, 16000.0f);
+
+        rec.SetEndpointerMode(EndpointerMode.LONG);
+
         using(Stream source = File.OpenRead("test.wav")) {
             byte[] buffer = new byte[4096];
             int bytesRead;
             while((bytesRead = source.Read(buffer, 0, buffer.Length)) > 0) {
                 float[] fbuffer = new float[bytesRead / 2];
                 for (int i = 0, n = 0; i < fbuffer.Length; i++, n+=2) {
-                    fbuffer[i] = (short)(buffer[n] | buffer[n+1] << 8);
+                    fbuffer[i] = BitConverter.ToInt16(buffer, n);
                 }
                 if (rec.AcceptWaveform(fbuffer, fbuffer.Length)) {
                     Console.WriteLine(rec.Result());

csharp/demo/VoskDemo.csproj

@@ -2,7 +2,7 @@
 
   <PropertyGroup>
     <OutputType>Exe</OutputType>
-    <TargetFramework>net5.0</TargetFramework>
+    <TargetFramework>net8.0</TargetFramework>
     <RootNamespace>VoskDemo</RootNamespace>
   </PropertyGroup>
 
@@ -11,7 +11,7 @@
   </PropertyGroup>
 
   <ItemGroup>
-    <PackageReference Include="Vosk" Version="0.3.30" />
+    <PackageReference Include="Vosk" Version="0.3.75" />
   </ItemGroup>
 
 </Project>

csharp/nuget/Vosk.csproj

@@ -0,0 +1,17 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+  <PropertyGroup>
+    <TargetFramework>net8.0</TargetFramework>
+    <ImplicitUsings>enable</ImplicitUsings>
+    <Nullable>enable</Nullable>
+    <PackageId>Vosk</PackageId>
+    <Version>0.3.75</Version>
+    <authors>Alpha Cephei Inc</authors>
+    <owners>Alpha Cephei Inc</owners>
+  </PropertyGroup>
+
+  <Target Name="CopyFiles" AfterTargets="Build">
+    <Copy SourceFiles="bin/Release/net8.0/Vosk.dll" DestinationFolder="lib/net8.0" />
+  </Target>
+
+</Project>

csharp/nuget/Vosk.nuspec

@@ -2,13 +2,14 @@
 <package>
   <metadata>
     <id>Vosk</id>
-    <version>0.3.30</version>
+    <version>0.3.75</version>
     <authors>Alpha Cephei Inc</authors>
     <owners>Alpha Cephei Inc</owners>
     <license type="expression">Apache-2.0</license>
     <projectUrl>https://alphacephei.com/vosk/</projectUrl>
     <requireLicenseAcceptance>false</requireLicenseAcceptance>
-    <description>Vosk is an offline open source speech recognition toolkit. It enables speech recognition models for 16 languages and dialects - English, Indian English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish, Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi.
+    <title>Vosk Speech Recognition Toolkit</title>
+    <description>Vosk is an offline open source speech recognition toolkit. It enables speech recognition models for 20+ languages and dialects - English, Indian English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish, Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi, Filipino, Ukrainian, Kazakh, Swedish, Japanese, Esperanto, Hindi, Czech. More to come.
 
 Vosk models are small (50 Mb) but provide continuous large vocabulary transcription, zero-latency response with streaming API, reconfigurable vocabulary and speaker identification.
 
@@ -18,13 +19,14 @@ Vosk supplies speech recognition for chatbots, smart home appliances, virtual as
 
 Vosk scales from small devices like Raspberry Pi or Android smartphone to big clusters.</description>
     <releaseNotes>See for details https://github.com/alphacep/vosk-api/releases</releaseNotes>
-    <copyright>Copyright 2020 Alpha Cephei Inc</copyright>
+    <repository type="git" url="https://github.com/alphacep/vosk-api.git" branch="master"/>
+    <copyright>Copyright 2020-2050 Alpha Cephei Inc</copyright>
     <tags>speech recognition voice stt asr speech-to-text ai offline privacy</tags>
     <dependencies>
-      <group targetFramework=".NETStandard2.0"/>
+      <group targetFramework="net8.0"/>
     </dependencies>
   </metadata>
   <files>
-    <file src="**" exclude="src/*.cs;build.sh;**/.keep-me;*.nupkg" />
+    <file src="**" exclude="bin/**;obj/**;build.sh;src/*.cs;*.nupkg;**/.keep-me" />
   </files>
 </package>

csharp/nuget/build.sh

@@ -1,2 +1,2 @@
-mcs -out:lib/netstandard2.0/Vosk.dll -target:library src/*.cs
-nuget pack
+rm -rf bin lib obj
+/home/shmyrev/local/dotnet/dotnet pack Vosk.csproj -p:NuspecFile=Vosk.nuspec -o .

csharp/nuget/build/Vosk.targets

@@ -2,7 +2,7 @@
   <ItemGroup>
     <NativeLibs Include="$(MSBuildThisFileDirectory)\lib\linux-x64\*.so" Condition="'$([MSBuild]::IsOsPlatform(Linux))'" />
     <NativeLibs Include="$(MSBuildThisFileDirectory)\lib\win-x64\*.dll" Condition="'$([MSBuild]::IsOsPlatform(Windows))'" />
-    <NativeLibs Include="$(MSBuildThisFileDirectory)\lib\osx-x64\*.dylib" Condition="'$([MSBuild]::IsOsPlatform(OSX))'" />
+    <NativeLibs Include="$(MSBuildThisFileDirectory)\lib\osx-universal\*.dylib" Condition="'$([MSBuild]::IsOsPlatform(OSX))'" />
     <None Include="@(NativeLibs)">
       <Link>%(FileName)%(Extension)</Link>
       <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>

csharp/nuget/src/BatchModel.cs

@@ -0,0 +1,55 @@
+using System;
+
+namespace Vosk
+{
+    public class BatchModel : global::System.IDisposable
+    {
+        private global::System.Runtime.InteropServices.HandleRef handle;
+
+        internal BatchModel(global::System.IntPtr cPtr)
+        {
+            handle = new global::System.Runtime.InteropServices.HandleRef(this, cPtr);
+        }
+
+        internal static global::System.Runtime.InteropServices.HandleRef getCPtr(BatchModel obj)
+        {
+            return (obj == null) ? new global::System.Runtime.InteropServices.HandleRef(null, global::System.IntPtr.Zero) : obj.handle;
+        }
+
+        ~BatchModel()
+        {
+            Dispose(false);
+        }
+
+        public void Dispose()
+        {
+            Dispose(true);
+            global::System.GC.SuppressFinalize(this);
+        }
+
+        protected virtual void Dispose(bool disposing)
+        {
+            lock (this)
+            {
+                if (handle.Handle != global::System.IntPtr.Zero)
+                {
+                    VoskPINVOKE.delete_BatchModel(handle);
+                    handle = new global::System.Runtime.InteropServices.HandleRef(null, global::System.IntPtr.Zero);
+                }
+            }
+        }
+
+        public BatchModel(string model_path) : this(VoskPINVOKE.new_BatchModel(model_path))
+        {
+        }
+
+        public void WaitForCompletion()
+        {
+            if (handle.Handle != global::System.IntPtr.Zero)
+            {
+                VoskPINVOKE.wait_BatchModel(handle);
+            }
+        }
+    }
+
+}

csharp/nuget/src/Model.cs

@@ -32,7 +32,7 @@ public class Model : global::System.IDisposable {
   public Model(string model_path) : this(VoskPINVOKE.new_Model(model_path)) {
   }
 
-  public int vosk_model_find_word(string word) {
+  public int FindWord(string word) {
     return VoskPINVOKE.Model_vosk_model_find_word(handle, word);
   }
 

csharp/nuget/src/VoskBatchRecognizer.cs

@@ -0,0 +1,88 @@
+using System;
+
+namespace Vosk
+{
+    public class VoskBatchRecognizer : System.IDisposable
+    {
+        private System.Runtime.InteropServices.HandleRef handle;
+
+        internal VoskBatchRecognizer(System.IntPtr cPtr)
+        {
+            handle = new System.Runtime.InteropServices.HandleRef(this, cPtr);
+        }
+
+        internal static System.Runtime.InteropServices.HandleRef getCPtr(VoskBatchRecognizer obj)
+        {
+            return (obj == null) ? new System.Runtime.InteropServices.HandleRef(null, System.IntPtr.Zero) : obj.handle;
+        }
+
+        ~VoskBatchRecognizer()
+        {
+            Dispose(false);
+        }
+
+        public void Dispose()
+        {
+            Dispose(true);
+            System.GC.SuppressFinalize(this);
+        }
+
+        protected virtual void Dispose(bool disposing)
+        {
+            lock (this)
+            {
+                if (handle.Handle != System.IntPtr.Zero)
+                {
+                    VoskPINVOKE.delete_VoskBatchRecognizer(handle);
+                    handle = new System.Runtime.InteropServices.HandleRef(null, System.IntPtr.Zero);
+                }
+            }
+        }
+
+        public VoskBatchRecognizer(BatchModel model, float sample_rate) : this(VoskPINVOKE.new_VoskBatchRecognizer(BatchModel.getCPtr(model), sample_rate))
+        {
+        }
+
+        public bool AcceptWaveform(byte[] data, int len)
+        {
+            return VoskPINVOKE.VoskBatchRecognizer_AcceptWaveform(handle, data, len);
+        }
+
+        private static string PtrToStringUTF8(System.IntPtr ptr)
+        {
+            int len = 0;
+            while (System.Runtime.InteropServices.Marshal.ReadByte(ptr, len) != 0)
+                len++;
+            byte[] array = new byte[len];
+            System.Runtime.InteropServices.Marshal.Copy(ptr, array, 0, len);
+            return System.Text.Encoding.UTF8.GetString(array);
+        }
+
+        public string FrontResult()
+        {
+            return PtrToStringUTF8(VoskPINVOKE.VoskBatchRecognizer_FrontResult(handle));
+        }
+
+        public string Result()
+        {
+            string result = PtrToStringUTF8(VoskPINVOKE.VoskBatchRecognizer_FrontResult(handle));
+            VoskPINVOKE.VoskBatchRecognizer_Pop(handle);
+            return result;
+        }
+
+        public int GetNumPendingChunks()
+        {
+            return VoskPINVOKE.VoskBatchRecognizer_GetPendingChunks(handle);
+        }
+
+        public void FinishStream()
+        {
+            VoskPINVOKE.VoskBatchRecognizer_FinishStream(handle);
+        }
+
+        public void SetNLSML(bool nlsml)
+        {
+            VoskPINVOKE.VoskBatchRecognizer_SetNLSML(handle, Convert.ToInt32(nlsml));
+        }
+    }
+}

csharp/nuget/src/VoskPINVOKE.cs

@@ -38,6 +38,9 @@ class VoskPINVOKE {
   [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_set_words")]
   public static extern void VoskRecognizer_SetWords(global::System.Runtime.InteropServices.HandleRef jarg1, int jarg2);
 
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_set_partial_words")]
+  public static extern void VoskRecognizer_SetPartialWords(global::System.Runtime.InteropServices.HandleRef jarg1, int jarg2);
+
   [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_set_spk_model")]
   public static extern void VoskRecognizer_SetSpkModel(global::System.Runtime.InteropServices.HandleRef jarg1, global::System.Runtime.InteropServices.HandleRef jarg2);
 
@@ -62,6 +65,12 @@ class VoskPINVOKE {
   [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_reset")]
   public static extern void VoskRecognizer_Reset(global::System.Runtime.InteropServices.HandleRef jarg1);
 
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_set_endpointer_mode")]
+  public static extern void VoskRecognizer_SetEndpointerMode(global::System.Runtime.InteropServices.HandleRef jarg1, int jarg2);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_set_endpointer_delays")]
+  public static extern void VoskRecognizer_SetEndpointerDelays(global::System.Runtime.InteropServices.HandleRef jarg1, float jarg2, float jarg3, float jarg4);
+
   [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_set_log_level")]
   public static extern void SetLogLevel(int jarg1);
 
@@ -70,6 +79,40 @@ class VoskPINVOKE {
 
   [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_gpu_thread_init")]
   public static extern void GpuThreadInit();
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_model_new")]
+  public static extern global::System.IntPtr new_BatchModel(string jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_model_free")]
+  public static extern void delete_BatchModel(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_model_wait")]
+  public static extern void wait_BatchModel(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_recognizer_new")]
+  public static extern global::System.IntPtr new_VoskBatchRecognizer(global::System.Runtime.InteropServices.HandleRef jarg1, float jarg2);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_recognizer_free")]
+  public static extern void delete_VoskBatchRecognizer(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_recognizer_accept_waveform")]
+  public static extern bool VoskBatchRecognizer_AcceptWaveform(global::System.Runtime.InteropServices.HandleRef jarg1, [global::System.Runtime.InteropServices.In, global::System.Runtime.InteropServices.MarshalAs(global::System.Runtime.InteropServices.UnmanagedType.LPArray)] byte[] jarg2, int jarg3);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_recognizer_set_nlsml")]
+  public static extern void VoskBatchRecognizer_SetNLSML(global::System.Runtime.InteropServices.HandleRef jarg1, int jarg2);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_recognizer_finish_stream")]
+  public static extern void VoskBatchRecognizer_FinishStream(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_recognizer_front_result")]
+  public static extern global::System.IntPtr VoskBatchRecognizer_FrontResult(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_recognizer_pop")]
+  public static extern void VoskBatchRecognizer_Pop(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint = "vosk_batch_recognizer_get_pending_chunks")]
+  public static extern int VoskBatchRecognizer_GetPendingChunks(global::System.Runtime.InteropServices.HandleRef jarg1);
+
 }
 
 }

csharp/nuget/src/VoskRecognizer.cs

@@ -1,5 +1,12 @@
 namespace Vosk {
 
+public enum EndpointerMode {
+    DEFAULT = 0,
+    SHORT = 1,
+    LONG = 2,
+    VERY_LONG = 3
+}
+
 public class VoskRecognizer : System.IDisposable {
   private System.Runtime.InteropServices.HandleRef handle;
 
@@ -46,6 +53,10 @@ public class VoskRecognizer : System.IDisposable {
     VoskPINVOKE.VoskRecognizer_SetWords(handle, words ? 1 : 0);
   }
 
+  public void SetPartialWords(bool partial_words) {
+    VoskPINVOKE.VoskRecognizer_SetPartialWords(handle, partial_words ? 1 : 0);
+  }
+
   public void SetSpkModel(SpkModel spk_model) {
     VoskPINVOKE.VoskRecognizer_SetSpkModel(handle, SpkModel.getCPtr(spk_model));
   }
@@ -87,6 +98,14 @@ public class VoskRecognizer : System.IDisposable {
     VoskPINVOKE.VoskRecognizer_Reset(handle);
   }
 
+  public void SetEndpointerMode(EndpointerMode mode) {
+    VoskPINVOKE.VoskRecognizer_SetEndpointerMode(handle, (int) mode);
+  }
+
+  public void SetEndpointerDelays(float t_start_max, float t_end, float t_max) {
+    VoskPINVOKE.VoskRecognizer_SetEndpointerDelays(handle, t_start_max, t_end, t_max);
+  }
+
 }
 
 }

go/COPYING

@@ -0,0 +1,176 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. this License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS

go/README.md

@@ -0,0 +1 @@
+See example subfolder for instructions how to use the module

go/batch.go

@@ -0,0 +1,99 @@
+package vosk
+
+// #cgo CPPFLAGS: -I ${SRCDIR}/../src
+// #cgo !windows LDFLAGS: -L ${SRCDIR}/../src -lvosk -ldl -lpthread
+// #cgo windows LDFLAGS: -L ${SRCDIR}/../src -lvosk -lpthread
+// #include <stdlib.h>
+// #include <vosk_api.h>
+import "C"
+import "unsafe"
+
+// VoskBatchModel contains a reference to the C VoskBatchModel
+type VoskBatchModel struct {
+	model *C.struct_VoskBatchModel
+}
+
+// NewBatchModel creates a new VoskBatchModel instance
+func NewBatchModel(modelPath string) (*VoskBatchModel, error) {
+	cmodelPath := C.CString(modelPath)
+	defer C.free(unsafe.Pointer(cmodelPath))
+	internal := C.vosk_batch_model_new(cmodelPath)
+	model := &VoskBatchModel{model: internal}
+	return model, nil
+}
+
+func (m *VoskBatchModel) Free() {
+	C.vosk_batch_model_free(m.model)
+}
+
+func (m *VoskBatchModel) Wait() {
+	C.vosk_batch_model_wait(m.model);
+}
+
+func freeBatchModel(model *VoskBatchModel) {
+	C.vosk_batch_model_free(model.model)
+}
+
+// VoskBatchRecognizer contains a reference to the C VoskBatchRecognizer
+type VoskBatchRecognizer struct {
+	rec *C.struct_VoskBatchRecognizer
+}
+
+func freeBatchRecognizer(recognizer *VoskBatchRecognizer) {
+	C.vosk_batch_recognizer_free(recognizer.rec)
+}
+
+func (r *VoskBatchRecognizer) Free() {
+	C.vosk_batch_recognizer_free(r.rec)
+}
+
+// NewBatchRecognizer creates a new VoskBatchRecognizer instance
+func NewBatchRecognizer(model *VoskBatchModel, sampleRate float64) (*VoskBatchRecognizer, error) {
+	internal := C.vosk_batch_recognizer_new(model.model, C.float(sampleRate))
+	rec := &VoskBatchRecognizer{rec: internal}
+	return rec, nil
+}
+
+// AcceptWaveform accepts and processes a new chunk of the voice data.
+func (r *VoskBatchRecognizer) AcceptWaveform(buffer []byte) {
+	cbuf := C.CBytes(buffer)
+	defer C.free(cbuf)
+	C.vosk_batch_recognizer_accept_waveform(r.rec, (*C.char)(cbuf), C.int(len(buffer)))
+}
+
+/** Set NLSML output
+ * @param nlsml - boolean value
+ */
+//void vosk_batch_recognizer_set_nlsml(VoskBatchRecognizer *recognizer, int nlsml);
+
+func (r *VoskBatchRecognizer) SetNlsml(nlsml int) {
+	C.vosk_batch_recognizer_set_nlsml(r.rec, C.int(nlsml))
+}
+
+/** Closes the stream */
+//void vosk_batch_recognizer_finish_stream(VoskBatchRecognizer *recognizer);
+
+func (r *VoskBatchRecognizer) FinishStream() {
+	C.vosk_batch_recognizer_finish_stream(r.rec)
+}
+
+/** Return results */
+//const char *vosk_batch_recognizer_front_result(VoskBatchRecognizer *recognizer);
+
+func (r *VoskBatchRecognizer) FrontResult() string {
+	return C.GoString(C.vosk_batch_recognizer_front_result(r.rec))
+}
+
+/** Release and free first retrieved result */
+//void vosk_batch_recognizer_pop(VoskBatchRecognizer *recognizer);
+
+func (r *VoskBatchRecognizer) Pop() {
+	C.vosk_batch_recognizer_pop(r.rec)
+}
+
+/** Get amount of pending chunks for more intelligent waiting */
+//int vosk_batch_recognizer_get_pending_chunks(VoskBatchRecognizer *recognizer);
+func (r *VoskBatchRecognizer) GetPendingChunks() int {
+	i := C.vosk_batch_recognizer_get_pending_chunks(r.rec)
+	return int(i)
+}

go/batch_example/README.md

@@ -0,0 +1,5 @@
+This example expects a `s16le` converted audio file and converts it to text in a
+manner that imitates the Python example of [test_gpu_batch.py](../python/example/test_gpu_batch.py).
+
+Note that the `libvosk.so` must be in the library path. This was successfully tested on
+Ubuntu 24.04 with Go 1.18, gcc-11, NVIDIA driver 570.172.08.

go/batch_example/test_batch.go

@@ -0,0 +1,54 @@
+package main
+
+import (
+	"flag"
+	"fmt"
+	"io"
+	"log"
+	"os"
+
+	vosk "github.com/alphacep/vosk-api/go"
+)
+
+func main() {
+	var filename string
+	flag.StringVar(&filename, "f", "", "file to transcribe")
+	flag.Parse()
+
+	vosk.GPUInit()
+
+	model, err := vosk.NewBatchModel("model")
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	rec, err := vosk.NewBatchRecognizer(model, 16000.0)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	file, err := os.Open(filename)
+	if err != nil {
+		panic(err)
+	}
+	defer file.Close()
+
+	buf := make([]byte, 8000)
+
+	for {
+		if _, err := file.Read(buf); err != nil {
+			if err != io.EOF {
+				log.Fatal(err)
+			}
+
+			break
+		}
+		rec.AcceptWaveform(buf)
+		model.Wait()
+		if rec.FrontResult() != "" {
+			fmt.Println(rec.FrontResult())
+			rec.Pop()
+		}
+	}
+	// Is this needed? rec.FinishStream()
+}

go/doc.go

@@ -0,0 +1,7 @@
+// Go bindings for Vosk speech recognition toolkit. Vosk is an offline
+// open source speech to text API for Android, iOS, Raspberry Pi and
+// servers. It enables speech recognition models for 18 languages and
+// dialects - English, Indian English, German, French, Spanish, Portuguese,
+// Chinese, Russian, Turkish, Vietnamese, Italian, Dutch, Catalan, Arabic,
+// Greek, Farsi, Filipino, Ukrainian.
+package vosk

go/example/README.md

@@ -0,0 +1,31 @@
+To try this package do the following steps:
+
+On Linux (we download library and set LD_LIBRARY_PATH)
+
+```
+git clone https://github.com/alphacep/vosk-api
+cd vosk-api/go/example
+wget https://github.com/alphacep/vosk-api/releases/download/v0.3.45/vosk-linux-x86_64-0.3.45.zip
+unzip vosk-linux-x86_64-0.3.45.zip
+wget https://alphacephei.com/vosk/models/vosk-model-small-en-us-0.15.zip
+unzip vosk-model-small-en-us-0.15.zip
+mv vosk-model-small-en-us-0.15 model
+cp ../../python/example/test.wav .
+VOSK_PATH=`pwd`/vosk-linux-x86_64-0.3.45 LD_LIBRARY_PATH=$VOSK_PATH CGO_CPPFLAGS="-I $VOSK_PATH" CGO_LDFLAGS="-L $VOSK_PATH" go run . -f test.wav
+```
+
+for Windows (we place DLLs in current folder where linker finds them):
+
+```
+git clone https://github.com/alphacep/vosk-api
+cd vosk-api/go/example
+wget https://github.com/alphacep/vosk-api/releases/download/v0.3.45/vosk-linux-x86_64-0.3.45.zip
+unzip vosk-linux-x86_64-0.3.45.zip
+cp vosk-linux-x86_64-0.3.45/*.dll .
+cp vosk-linux-x86_64-0.3.45/*.h .
+wget https://alphacephei.com/vosk/models/vosk-model-small-en-us-0.15.zip
+unzip vosk-model-small-en-us-0.15.zip
+mv vosk-model-small-en-us-0.15 model
+cp ../../python/example/test.wav .
+VOSK_PATH=`pwd` LD_LIBRARY_PATH=$VOSK_PATH CGO_CPPFLAGS="-I $VOSK_PATH" CGO_LDFLAGS="-L $VOSK_PATH -lvosk -lpthread -dl" go run . -f test.wav
+```

go/example/doc.go

@@ -0,0 +1,2 @@
+// Example package for Vosk Go bindings.
+package main

go/example/main.go

@@ -1,36 +0,0 @@
-package main
-
-import (
-    "flag"
-    "os"
-    ".."
-)
-
-func main() {
-    var filename string
-    flag.StringVar(&filename, "f", "", "file to transcribe")
-    flag.Parse()
-    model, err := vosk.NewModel("model")
-    rec, err := vosk.NewRecognizer(model)
-
-    file, err := os.Open(filename)
-    if err != nil {
-        panic(err)
-    }
-    defer file.Close()
-
-    fileinfo, err := file.Stat()
-    if err != nil {
-        panic(err)
-    }
-
-    filesize := fileinfo.Size()
-    buffer := make([]byte, filesize)
-
-    _, err = file.Read(buffer)
-    if err != nil {
-        panic(err)
-    }
-
-    println(vosk.VoskFinalResult(rec, buffer))
-}

go/example/test_simple.go

@@ -0,0 +1,63 @@
+package main
+
+import (
+	"encoding/json"
+	"flag"
+	"fmt"
+	"io"
+	"log"
+	"os"
+
+	vosk "github.com/alphacep/vosk-api/go"
+)
+
+func main() {
+	var filename string
+	flag.StringVar(&filename, "f", "", "file to transcribe")
+	flag.Parse()
+
+	vosk.GPUInit()
+
+	model, err := vosk.NewModel("model")
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	// we can check if word is in the vocabulary
+	// fmt.Println(model.FindWord("air"))
+
+	sampleRate := 16000.0
+	rec, err := vosk.NewRecognizer(model, sampleRate)
+	if err != nil {
+		log.Fatal(err)
+	}
+	rec.SetWords(1)
+
+	file, err := os.Open(filename)
+	if err != nil {
+		panic(err)
+	}
+	defer file.Close()
+
+	buf := make([]byte, 4096)
+
+	for {
+		_, err := file.Read(buf)
+		if err != nil {
+			if err != io.EOF {
+				log.Fatal(err)
+			}
+
+			break
+		}
+
+		if rec.AcceptWaveform(buf) != 0 {
+			fmt.Println(rec.Result())
+		}
+	}
+
+	// Unmarshal example for final result
+	var jres map[string]interface{}
+	json.Unmarshal([]byte(rec.FinalResult()), &jres)
+	fmt.Println(jres["text"])
+}

go/go.mod

@@ -1,3 +1,7 @@
-module alphacephei.com/vosk
+module github.com/alphacep/vosk-api/go
 
 go 1.16
+
+replace (
+    github.com/alphacep/vosk-api/go => ./
+)

go/vosk.go

@@ -1,62 +1,180 @@
 package vosk
 
 // #cgo CPPFLAGS: -I ${SRCDIR}/../src
-// #cgo LDFLAGS: -L ${SRCDIR}/../src -lvosk -ldl -lpthread
+// #cgo !windows LDFLAGS: -L ${SRCDIR}/../src -lvosk -ldl -lpthread
+// #cgo windows LDFLAGS: -L ${SRCDIR}/../src -lvosk -lpthread
 // #include <stdlib.h>
 // #include <vosk_api.h>
 import "C"
+import "unsafe"
 
 // VoskModel contains a reference to the C VoskModel
 type VoskModel struct {
-    model *C.struct_VoskModel
-}
-
-// VoskSpkModel contains a reference to the C VoskSpkModel
-type VoskSpkModel struct {
-    spkModel *C.struct_VoskSpkModel
-}
-
-// VoskRecognizer contains a reference to the C VoskRecognizer
-type VoskRecognizer struct {
-    rec *C.struct_VoskRecognizer
-}
-
-func VoskFinalResult(recognizer *VoskRecognizer, buffer []byte) string {
-    cbuf := C.CBytes(buffer)
-    defer C.free(cbuf)
-    _ = C.vosk_recognizer_accept_waveform(recognizer.rec, (*C.char)(cbuf), C.int(len(buffer)))
-    result := C.GoString(C.vosk_recognizer_final_result(recognizer.rec))
-    return result
+	model *C.struct_VoskModel
 }
 
 // NewModel creates a new VoskModel instance
 func NewModel(modelPath string) (*VoskModel, error) {
-    var internal *C.struct_VoskModel
-    internal = C.vosk_model_new(C.CString(modelPath))
-    model := &VoskModel{model: internal}
-    return model, nil
+	cmodelPath := C.CString(modelPath)
+	defer C.free(unsafe.Pointer(cmodelPath))
+	internal := C.vosk_model_new(cmodelPath)
+	model := &VoskModel{model: internal}
+	return model, nil
 }
 
-// NewRecognizer creates a new VoskRecognizer instance
-func NewRecognizer(model *VoskModel) (*VoskRecognizer, error) {
-    var internal *C.struct_VoskRecognizer
-    internal = C.vosk_recognizer_new(model.model, 16000.0)
-    rec := &VoskRecognizer{rec: internal}
-    return rec, nil
+func (m *VoskModel) Free() {
+	C.vosk_model_free(m.model)
 }
 
 func freeModel(model *VoskModel) {
-    C.vosk_model_free(model.model)
+	C.vosk_model_free(model.model)
 }
 
-func freeRecognizer(recognizer *VoskRecognizer) {
-    C.vosk_recognizer_free(recognizer.rec)
+// FindWord checks if a word can be recognized by the model.
+// Returns the word symbol if the word exists inside the model or
+// -1 otherwise.
+func (m *VoskModel) FindWord(word string) int {
+	cstr := C.CString(word)
+	defer C.free(unsafe.Pointer(cstr))
+	i := C.vosk_model_find_word(m.model, cstr)
+	return int(i)
+}
+
+// VoskSpkModel contains a reference to the C VoskSpkModel
+type VoskSpkModel struct {
+	spkModel *C.struct_VoskSpkModel
 }
 
 // NewSpkModel creates a new VoskSpkModel instance
 func NewSpkModel(spkModelPath string) (*VoskSpkModel, error) {
-    var internal *C.struct_VoskSpkModel
-    internal = C.vosk_spk_model_new(C.CString(spkModelPath))
-    spkModel := &VoskSpkModel{spkModel: internal}
-    return spkModel, nil
+	cspkModelPath := C.CString(spkModelPath)
+	defer C.free(unsafe.Pointer(cspkModelPath))
+	internal := C.vosk_spk_model_new(cspkModelPath)
+	spkModel := &VoskSpkModel{spkModel: internal}
+	return spkModel, nil
+}
+
+func freeSpkModel(model *VoskSpkModel) {
+	C.vosk_spk_model_free(model.spkModel)
+}
+
+func(s *VoskSpkModel) Free() {
+	C.vosk_spk_model_free(s.spkModel)
+}
+
+// VoskRecognizer contains a reference to the C VoskRecognizer
+type VoskRecognizer struct {
+	rec *C.struct_VoskRecognizer
+}
+
+func freeRecognizer(recognizer *VoskRecognizer) {
+	C.vosk_recognizer_free(recognizer.rec)
+}
+
+func (r *VoskRecognizer) Free() {
+	C.vosk_recognizer_free(r.rec)
+}
+
+// NewRecognizer creates a new VoskRecognizer instance
+func NewRecognizer(model *VoskModel, sampleRate float64) (*VoskRecognizer, error) {
+	internal := C.vosk_recognizer_new(model.model, C.float(sampleRate))
+	rec := &VoskRecognizer{rec: internal}
+	return rec, nil
+}
+
+// NewRecognizerSpk creates a new VoskRecognizer instance with a speaker model.
+func NewRecognizerSpk(model *VoskModel, sampleRate float64, spkModel *VoskSpkModel) (*VoskRecognizer, error) {
+	internal := C.vosk_recognizer_new_spk(model.model, C.float(sampleRate), spkModel.spkModel)
+	rec := &VoskRecognizer{rec: internal}
+	return rec, nil
+}
+
+// NewRecognizerGrm creates a new VoskRecognizer instance with the phrase list.
+func NewRecognizerGrm(model *VoskModel, sampleRate float64, grammar string) (*VoskRecognizer, error) {
+	cgrammar := C.CString(grammar)
+	defer C.free(unsafe.Pointer(cgrammar))
+	internal := C.vosk_recognizer_new_grm(model.model, C.float(sampleRate), cgrammar)
+	rec := &VoskRecognizer{rec: internal}
+	return rec, nil
+}
+
+// SetSpkModel adds a speaker model to an already initialized recognizer.
+func (r *VoskRecognizer) SetSpkModel(spkModel *VoskSpkModel) {
+	C.vosk_recognizer_set_spk_model(r.rec, spkModel.spkModel)
+}
+
+// SetGrm sets which phrases to recognize on an already initialized recognizer.
+func (r *VoskRecognizer) SetGrm(grammar string) {
+	cgrammar := C.CString(grammar)
+	defer C.free(unsafe.Pointer(cgrammar))
+	C.vosk_recognizer_set_grm(r.rec, cgrammar)
+}
+
+// SetMaxAlternatives configures the recognizer to output n-best results.
+func (r *VoskRecognizer) SetMaxAlternatives(maxAlternatives int) {
+	C.vosk_recognizer_set_max_alternatives(r.rec, C.int(maxAlternatives))
+}
+
+// SetWords enables words with times in the ouput.
+func (r *VoskRecognizer) SetWords(words int) {
+	C.vosk_recognizer_set_words(r.rec, C.int(words))
+}
+
+// SetPartialWords enables words with times in the partial ouput.
+func (r *VoskRecognizer) SetPartialWords(words int) {
+	C.vosk_recognizer_set_partial_words(r.rec, C.int(words))
+}
+
+// SetEndpointerDelays sets the recognition timeouts, where startMax
+// is the timeout for stopping recognition in case of initial silence
+// (usually around 5), end is the timeout for stopping recognition
+// in milliseconds after we recognized something (usually around 0.5-1.0),
+// and max is the timeout for forcing utterance end in milliseconds
+// (usually around 20-30).
+func (r *VoskRecognizer) SetEndpointerDelays(startMax, end, max float64) {
+	C.vosk_recognizer_set_endpointer_delays(r.rec, C.float(startMax), C.float(end), C.float(max))
+}
+
+// AcceptWaveform accepts and processes a new chunk of the voice data.
+func (r *VoskRecognizer) AcceptWaveform(buffer []byte) int {
+	cbuf := C.CBytes(buffer)
+	defer C.free(cbuf)
+	i := C.vosk_recognizer_accept_waveform(r.rec, (*C.char)(cbuf), C.int(len(buffer)))
+	return int(i)
+}
+
+// Result returns a speech recognition result.
+func (r *VoskRecognizer) Result() string {
+	return C.GoString(C.vosk_recognizer_result(r.rec))
+}
+
+// PartialResult returns a partial speech recognition result.
+func (r *VoskRecognizer) PartialResult() string {
+	return C.GoString(C.vosk_recognizer_partial_result(r.rec))
+}
+
+// FinalResult returns a speech recognition result. Same as result, but doesn't wait
+// for silence.
+func (r *VoskRecognizer) FinalResult() string {
+	return C.GoString(C.vosk_recognizer_final_result(r.rec))
+}
+
+// Reset resets the recognizer.
+func (r *VoskRecognizer) Reset() {
+	C.vosk_recognizer_reset(r.rec)
+}
+
+// SetLogLevel sets the log level for Kaldi messages.
+func SetLogLevel(logLevel int) {
+	C.vosk_set_log_level(C.int(logLevel))
+}
+
+// GPUInit automatically selects a CUDA device and allows multithreading.
+func GPUInit() {
+	C.vosk_gpu_init()
+}
+
+// GPUThreadInit inits CUDA device in a multi-threaded environment.
+func GPUThreadInit() {
+	C.vosk_gpu_thread_init()
 }

ios/VoskApiTest.xcodeproj/project.pbxproj

@@ -13,10 +13,10 @@
 		92375229240C550B00DD6076 /* Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = 92375228240C550B00DD6076 /* Assets.xcassets */; };
 		9237522C240C550B00DD6076 /* LaunchScreen.storyboard in Resources */ = {isa = PBXBuildFile; fileRef = 9237522A240C550B00DD6076 /* LaunchScreen.storyboard */; };
 		92375234240C558900DD6076 /* Vosk.swift in Sources */ = {isa = PBXBuildFile; fileRef = 92375233240C558900DD6076 /* Vosk.swift */; };
-		9237523C240C642000DD6076 /* libkaldiwrap.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 9237523A240C642000DD6076 /* libkaldiwrap.a */; };
 		92375244240C6DAF00DD6076 /* Accelerate.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 92375243240C6DAF00DD6076 /* Accelerate.framework */; };
-		92375246240C6DC900DD6076 /* libstdc++.tbd in Frameworks */ = {isa = PBXBuildFile; fileRef = 92375245240C6DC900DD6076 /* libstdc++.tbd */; };
 		92375274240C6F1E00DD6076 /* 10001-90210-01803.wav in Resources */ = {isa = PBXBuildFile; fileRef = 92375256240C6E3D00DD6076 /* 10001-90210-01803.wav */; };
+		925527A9273C492C00FFD9CC /* libvosk.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 925527A8273C492C00FFD9CC /* libvosk.a */; };
+		92833003273C466E00058B52 /* libc++.tbd in Frameworks */ = {isa = PBXBuildFile; fileRef = 92833002273C466E00058B52 /* libc++.tbd */; };
 		92BACED125BE125A00B5CC93 /* vosk-model-small-en-us-0.15 in Resources */ = {isa = PBXBuildFile; fileRef = 928CC50C25BE124400490481 /* vosk-model-small-en-us-0.15 */; };
 		92D6B8D325BDFEAC007FF08D /* VoskModel.swift in Sources */ = {isa = PBXBuildFile; fileRef = 92D6B8D225BDFEAC007FF08D /* VoskModel.swift */; };
 		92D86BD6253F823F0040D53F /* vosk-model-spk-0.4 in Resources */ = {isa = PBXBuildFile; fileRef = 92D86BD4253F823F0040D53F /* vosk-model-spk-0.4 */; };
@@ -31,10 +31,10 @@
 		9237522B240C550B00DD6076 /* Base */ = {isa = PBXFileReference; lastKnownFileType = file.storyboard; name = Base; path = Base.lproj/LaunchScreen.storyboard; sourceTree = "<group>"; };
 		9237522D240C550B00DD6076 /* Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
 		92375233240C558900DD6076 /* Vosk.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = Vosk.swift; sourceTree = "<group>"; };
-		9237523A240C642000DD6076 /* libkaldiwrap.a */ = {isa = PBXFileReference; lastKnownFileType = archive.ar; path = libkaldiwrap.a; sourceTree = "<group>"; };
 		92375243240C6DAF00DD6076 /* Accelerate.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = Accelerate.framework; path = System/Library/Frameworks/Accelerate.framework; sourceTree = SDKROOT; };
-		92375245240C6DC900DD6076 /* libstdc++.tbd */ = {isa = PBXFileReference; lastKnownFileType = "sourcecode.text-based-dylib-definition"; name = "libstdc++.tbd"; path = "usr/lib/libstdc++.tbd"; sourceTree = SDKROOT; };
 		92375256240C6E3D00DD6076 /* 10001-90210-01803.wav */ = {isa = PBXFileReference; lastKnownFileType = audio.wav; path = "10001-90210-01803.wav"; sourceTree = "<group>"; };
+		925527A8273C492C00FFD9CC /* libvosk.a */ = {isa = PBXFileReference; lastKnownFileType = archive.ar; path = libvosk.a; sourceTree = "<group>"; };
+		92833002273C466E00058B52 /* libc++.tbd */ = {isa = PBXFileReference; lastKnownFileType = "sourcecode.text-based-dylib-definition"; name = "libc++.tbd"; path = "usr/lib/libc++.tbd"; sourceTree = SDKROOT; };
 		928CC50C25BE124400490481 /* vosk-model-small-en-us-0.15 */ = {isa = PBXFileReference; lastKnownFileType = folder; name = "vosk-model-small-en-us-0.15"; path = "/Users/shmyrev/Documents/IOS/VoskApiTest/VoskApiTest/Vosk/vosk-model-small-en-us-0.15"; sourceTree = "<absolute>"; };
 		92AA22AD244CDD1200DA464B /* vosk_api.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = vosk_api.h; sourceTree = "<group>"; };
 		92AA22AE244CDD5200DA464B /* bridging.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = bridging.h; sourceTree = "<group>"; };
@@ -47,9 +47,9 @@
 			isa = PBXFrameworksBuildPhase;
 			buildActionMask = 2147483647;
 			files = (
-				92375246240C6DC900DD6076 /* libstdc++.tbd in Frameworks */,
+				92833003273C466E00058B52 /* libc++.tbd in Frameworks */,
 				92375244240C6DAF00DD6076 /* Accelerate.framework in Frameworks */,
-				9237523C240C642000DD6076 /* libkaldiwrap.a in Frameworks */,
+				925527A9273C492C00FFD9CC /* libvosk.a in Frameworks */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
 		};
@@ -93,11 +93,11 @@
 		92375239240C642000DD6076 /* Vosk */ = {
 			isa = PBXGroup;
 			children = (
-				928CC50C25BE124400490481 /* vosk-model-small-en-us-0.15 */,
 				92D86BD4253F823F0040D53F /* vosk-model-spk-0.4 */,
-				92375256240C6E3D00DD6076 /* 10001-90210-01803.wav */,
+				928CC50C25BE124400490481 /* vosk-model-small-en-us-0.15 */,
+				925527A8273C492C00FFD9CC /* libvosk.a */,
 				92AA22AD244CDD1200DA464B /* vosk_api.h */,
-				9237523A240C642000DD6076 /* libkaldiwrap.a */,
+				92375256240C6E3D00DD6076 /* 10001-90210-01803.wav */,
 			);
 			name = Vosk;
 			path = VoskApiTest/Vosk;
@@ -106,7 +106,7 @@
 		92375242240C6DAF00DD6076 /* Frameworks */ = {
 			isa = PBXGroup;
 			children = (
-				92375245240C6DC900DD6076 /* libstdc++.tbd */,
+				92833002273C466E00058B52 /* libc++.tbd */,
 				92375243240C6DAF00DD6076 /* Accelerate.framework */,
 			);
 			name = Frameworks;
@@ -145,7 +145,7 @@
 					9237521D240C550B00DD6076 = {
 						CreatedOnToolsVersion = 8.3.2;
 						LastSwiftMigration = 0920;
-						ProvisioningStyle = Automatic;
+						ProvisioningStyle = Manual;
 					};
 				};
 			};
@@ -223,7 +223,6 @@
 				ALWAYS_SEARCH_USER_PATHS = NO;
 				CLANG_ANALYZER_NONNULL = YES;
 				CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
-				CLANG_CXX_LANGUAGE_STANDARD = "gnu++0x";
 				CLANG_CXX_LIBRARY = "libc++";
 				CLANG_ENABLE_MODULES = YES;
 				CLANG_ENABLE_OBJC_ARC = YES;
@@ -245,7 +244,7 @@
 				CLANG_WARN_SUSPICIOUS_MOVE = YES;
 				CLANG_WARN_UNREACHABLE_CODE = YES;
 				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
-				"CODE_SIGN_IDENTITY[sdk=iphoneos*]" = "iPhone Developer";
+				"CODE_SIGN_IDENTITY[sdk=iphoneos*]" = "";
 				COPY_PHASE_STRIP = NO;
 				DEBUG_INFORMATION_FORMAT = dwarf;
 				ENABLE_STRICT_OBJC_MSGSEND = YES;
@@ -282,7 +281,6 @@
 				ALWAYS_SEARCH_USER_PATHS = NO;
 				CLANG_ANALYZER_NONNULL = YES;
 				CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
-				CLANG_CXX_LANGUAGE_STANDARD = "gnu++0x";
 				CLANG_CXX_LIBRARY = "libc++";
 				CLANG_ENABLE_MODULES = YES;
 				CLANG_ENABLE_OBJC_ARC = YES;
@@ -304,7 +302,7 @@
 				CLANG_WARN_SUSPICIOUS_MOVE = YES;
 				CLANG_WARN_UNREACHABLE_CODE = YES;
 				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
-				"CODE_SIGN_IDENTITY[sdk=iphoneos*]" = "iPhone Developer";
+				"CODE_SIGN_IDENTITY[sdk=iphoneos*]" = "";
 				COPY_PHASE_STRIP = NO;
 				DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
 				ENABLE_NS_ASSERTIONS = NO;
@@ -319,6 +317,7 @@
 				GCC_WARN_UNUSED_VARIABLE = YES;
 				IPHONEOS_DEPLOYMENT_TARGET = 10.3;
 				MTL_ENABLE_DEBUG_INFO = NO;
+				ONLY_ACTIVE_ARCH = YES;
 				SDKROOT = iphoneos;
 				SWIFT_OBJC_BRIDGING_HEADER = bridging.h;
 				SWIFT_OPTIMIZATION_LEVEL = "-Owholemodule";
@@ -333,7 +332,10 @@
 			buildSettings = {
 				ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
 				CLANG_ENABLE_MODULES = YES;
-				ENABLE_BITCODE = YES;
+				"CODE_SIGN_IDENTITY[sdk=iphoneos*]" = "iPhone Developer";
+				CODE_SIGN_STYLE = Manual;
+				DEVELOPMENT_TEAM = "";
+				ENABLE_BITCODE = NO;
 				INFOPLIST_FILE = VoskApiTest/Info.plist;
 				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
 				LIBRARY_SEARCH_PATHS = (
@@ -342,11 +344,13 @@
 				);
 				PRODUCT_BUNDLE_IDENTIFIER = com.alphacephei.VoskApiTest;
 				PRODUCT_NAME = "$(TARGET_NAME)";
+				PROVISIONING_PROFILE_SPECIFIER = "";
 				SWIFT_INSTALL_OBJC_HEADER = YES;
 				SWIFT_OBJC_BRIDGING_HEADER = VoskApiTest/bridging.h;
 				SWIFT_OPTIMIZATION_LEVEL = "-Onone";
 				SWIFT_SWIFT3_OBJC_INFERENCE = Default;
 				SWIFT_VERSION = 4.0;
+				TARGETED_DEVICE_FAMILY = "1,2";
 			};
 			name = Debug;
 		};
@@ -355,7 +359,10 @@
 			buildSettings = {
 				ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
 				CLANG_ENABLE_MODULES = YES;
-				ENABLE_BITCODE = YES;
+				"CODE_SIGN_IDENTITY[sdk=iphoneos*]" = "iPhone Developer";
+				CODE_SIGN_STYLE = Manual;
+				DEVELOPMENT_TEAM = "";
+				ENABLE_BITCODE = NO;
 				INFOPLIST_FILE = VoskApiTest/Info.plist;
 				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
 				LIBRARY_SEARCH_PATHS = (
@@ -364,10 +371,12 @@
 				);
 				PRODUCT_BUNDLE_IDENTIFIER = com.alphacephei.VoskApiTest;
 				PRODUCT_NAME = "$(TARGET_NAME)";
+				PROVISIONING_PROFILE_SPECIFIER = "";
 				SWIFT_INSTALL_OBJC_HEADER = YES;
 				SWIFT_OBJC_BRIDGING_HEADER = VoskApiTest/bridging.h;
 				SWIFT_SWIFT3_OBJC_INFERENCE = Default;
 				SWIFT_VERSION = 4.0;
+				TARGETED_DEVICE_FAMILY = "1,2";
 			};
 			name = Release;
 		};

ios/VoskApiTest.xcodeproj/project.xcworkspace/contents.xcworkspacedata

@@ -1,9 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
 <Workspace
    version = "1.0">
-   <FileRef
-      location = "group:/Users/shmyrev/Documents/IOS/VoskApiTest/VoskApiTest/Vosk/vosk-model-small-en-us-0.15">
-   </FileRef>
    <FileRef
       location = "self:">
    </FileRef>

ios/VoskApiTest/Vosk.swift

@@ -14,7 +14,7 @@ public final class Vosk {
     var recognizer : OpaquePointer!
 
     init(model: VoskModel, sampleRate: Float) {
-        recognizer = vosk_recognizer_new_spk(model.model, model.spkModel, sampleRate)
+        recognizer = vosk_recognizer_new_spk(model.model, sampleRate, model.spkModel)
     }
     
     deinit {

ios/VoskApiTest/Vosk/vosk_api.h

@@ -1,4 +1,4 @@
-// Copyright 2020 Alpha Cephei Inc.
+// Copyright 2020-2021 Alpha Cephei Inc.
 //
 // Licensed under the Apache License, Version 2.0 (the "License");
 // you may not use this file except in compliance with the License.
@@ -43,7 +43,7 @@ typedef struct VoskRecognizer VoskRecognizer;
 /** Loads model data from the file and returns the model object
  *
  * @param model_path: the path of the model on the filesystem
- @ @returns model object */
+ * @returns model object or NULL if problem occured */
 VoskModel *vosk_model_new(const char *model_path);
 
 
@@ -55,10 +55,18 @@ VoskModel *vosk_model_new(const char *model_path);
 void vosk_model_free(VoskModel *model);
 
 
+/** Check if a word can be recognized by the model
+ * @param word: the word
+ * @returns the word symbol if @param word exists inside the model
+ * or -1 otherwise.
+ * Reminding that word symbol 0 is for <epsilon> */
+int vosk_model_find_word(VoskModel *model, const char *word);
+
+
 /** Loads speaker model data from the file and returns the model object
  *
  * @param model_path: the path of the model on the filesystem
- * @returns model object */
+ * @returns model object or NULL if problem occured */
 VoskSpkModel *vosk_spk_model_new(const char *model_path);
 
 
@@ -71,9 +79,13 @@ void vosk_spk_model_free(VoskSpkModel *model);
 
 /** Creates the recognizer object
  *
- *  The recognizers process the speech and return text using shared model data 
- *  @param sample_rate The sample rate of the audio you going to feed into the recognizer
- *  @returns recognizer object */
+ *  The recognizers process the speech and return text using shared model data
+ *  @param model       VoskModel containing static data for recognizer. Model can be
+ *                     shared across recognizers, even running in different threads.
+ *  @param sample_rate The sample rate of the audio you going to feed into the recognizer.
+ *                     Make sure this rate matches the audio content, it is a common
+ *                     issue causing accuracy problems.
+ *  @returns recognizer object or NULL if problem occured */
 VoskRecognizer *vosk_recognizer_new(VoskModel *model, float sample_rate);
 
 
@@ -82,10 +94,14 @@ VoskRecognizer *vosk_recognizer_new(VoskModel *model, float sample_rate);
  *  With the speaker recognition mode the recognizer not just recognize
  *  text but also return speaker vectors one can use for speaker identification
  *
+ *  @param model       VoskModel containing static data for recognizer. Model can be
+ *                     shared across recognizers, even running in different threads.
+ *  @param sample_rate The sample rate of the audio you going to feed into the recognizer.
+ *                     Make sure this rate matches the audio content, it is a common
+ *                     issue causing accuracy problems.
  *  @param spk_model speaker model for speaker identification
- *  @param sample_rate The sample rate of the audio you going to feed into the recognizer
- *  @returns recognizer object */
-VoskRecognizer *vosk_recognizer_new_spk(VoskModel *model, VoskSpkModel *spk_model, float sample_rate);
+ *  @returns recognizer object or NULL if problem occured */
+VoskRecognizer *vosk_recognizer_new_spk(VoskModel *model, float sample_rate, VoskSpkModel *spk_model);
 
 
 /** Creates the recognizer object with the phrase list
@@ -98,42 +114,46 @@ VoskRecognizer *vosk_recognizer_new_spk(VoskModel *model, VoskSpkModel *spk_mode
  *  Only recognizers with lookahead models support this type of quick configuration.
  *  Precompiled HCLG graph models are not supported.
  *
- *  @param sample_rate The sample rate of the audio you going to feed into the recognizer
+ *  @param model       VoskModel containing static data for recognizer. Model can be
+ *                     shared across recognizers, even running in different threads.
+ *  @param sample_rate The sample rate of the audio you going to feed into the recognizer.
+ *                     Make sure this rate matches the audio content, it is a common
+ *                     issue causing accuracy problems.
  *  @param grammar The string with the list of phrases to recognize as JSON array of strings,
  *                 for example "["one two three four five", "[unk]"]".
  *
- *  @returns recognizer object */
+ *  @returns recognizer object or NULL if problem occured */
 VoskRecognizer *vosk_recognizer_new_grm(VoskModel *model, float sample_rate, const char *grammar);
 
 
-/** Accept voice data
+/** Adds speaker model to already initialized recognizer
  *
- *  accept and process new chunk of voice data
+ * Can add speaker recognition model to already created recognizer. Helps to initialize
+ * speaker recognition for grammar-based recognizer.
  *
- *  @param data - audio data in PCM 16-bit mono format
- *  @param length - length of the audio data
- *  @returns true if silence is occured and you can retrieve a new utterance with result method */
-int vosk_recognizer_accept_waveform(VoskRecognizer *recognizer, const char *data, int length);
+ * @param spk_model Speaker recognition model */
+void vosk_recognizer_set_spk_model(VoskRecognizer *recognizer, VoskSpkModel *spk_model);
 
 
-/** Same as above but the version with the short data for language bindings where you have
- *  audio as array of shorts */
-int vosk_recognizer_accept_waveform_s(VoskRecognizer *recognizer, const short *data, int length);
-
-
-/** Same as above but the version with the float data for language bindings where you have
- *  audio as array of floats */
-int vosk_recognizer_accept_waveform_f(VoskRecognizer *recognizer, const float *data, int length);
-
-
-/** Returns speech recognition result
- *
- * @returns the result in JSON format which contains decoded line, decoded
- *          words, times in seconds and confidences. You can parse this result
- *          with any json parser
+/** Configures recognizer to output n-best results
+ *
+ * <pre>
+ *   {
+ *      "alternatives": [
+ *          { "text": "one two three four five", "confidence": 0.97 },
+ *          { "text": "one two three for five", "confidence": 0.03 },
+ *      ]
+ *   }
+ * </pre>
+ *
+ * @param max_alternatives - maximum alternatives to return from recognition results
+ */
+void vosk_recognizer_set_max_alternatives(VoskRecognizer *recognizer, int max_alternatives);
+
+
+/** Enables words with times in the output
  *
  * <pre>
- * {
  *   "result" : [{
  *       "conf" : 1.000000,
  *       "end" : 1.110000,
@@ -156,13 +176,54 @@ int vosk_recognizer_accept_waveform_f(VoskRecognizer *recognizer, const float *d
  *       "word" : "zero"
  *     }, {
  *       "conf" : 1.000000,
- *      "end" : 2.610000,
+ *       "end" : 2.610000,
  *       "start" : 2.340000,
  *       "word" : "one"
  *     }],
- *   "text" : "what zero zero zero one"
+ * </pre>
+ *
+ * @param words - boolean value
+ */
+void vosk_recognizer_set_words(VoskRecognizer *recognizer, int words);
+
+
+/** Accept voice data
+ *
+ *  accept and process new chunk of voice data
+ *
+ *  @param data - audio data in PCM 16-bit mono format
+ *  @param length - length of the audio data
+ *  @returns 1 if silence is occured and you can retrieve a new utterance with result method 
+ *           0 if decoding continues
+ *           -1 if exception occured */
+int vosk_recognizer_accept_waveform(VoskRecognizer *recognizer, const char *data, int length);
+
+
+/** Same as above but the version with the short data for language bindings where you have
+ *  audio as array of shorts */
+int vosk_recognizer_accept_waveform_s(VoskRecognizer *recognizer, const short *data, int length);
+
+
+/** Same as above but the version with the float data for language bindings where you have
+ *  audio as array of floats */
+int vosk_recognizer_accept_waveform_f(VoskRecognizer *recognizer, const float *data, int length);
+
+
+/** Returns speech recognition result
+ *
+ * @returns the result in JSON format which contains decoded line, decoded
+ *          words, times in seconds and confidences. You can parse this result
+ *          with any json parser
+ *
+ * <pre>
+ *  {
+ *    "text" : "what zero zero zero one"
  *  }
  * </pre>
+ *
+ * If alternatives enabled it returns result with alternatives, see also vosk_recognizer_set_alternatives().
+ *
+ * If word times enabled returns word time, see also vosk_recognizer_set_word_times().
  */
 const char *vosk_recognizer_result(VoskRecognizer *recognizer);
 
@@ -174,7 +235,7 @@ const char *vosk_recognizer_result(VoskRecognizer *recognizer);
  *
  * <pre>
  * {
- *  "partial" : "cyril one eight zero"
+ *    "partial" : "cyril one eight zero"
  * }
  * </pre>
  */
@@ -190,6 +251,12 @@ const char *vosk_recognizer_partial_result(VoskRecognizer *recognizer);
 const char *vosk_recognizer_final_result(VoskRecognizer *recognizer);
 
 
+/** Resets the recognizer
+ *
+ *  Resets current results so the recognition can continue from scratch */
+void vosk_recognizer_reset(VoskRecognizer *recognizer);
+
+
 /** Releases recognizer object
  *
  *  Underlying model is also unreferenced and if needed released */
@@ -204,6 +271,20 @@ void vosk_recognizer_free(VoskRecognizer *recognizer);
  */
 void vosk_set_log_level(int log_level);
 
+/**
+ *  Init, automatically select a CUDA device and allow multithreading.
+ *  Must be called once from the main thread.
+ *  Has no effect if HAVE_CUDA flag is not set.
+ */
+void vosk_gpu_init();
+
+/**
+ *  Init CUDA device in a multi-threaded environment.
+ *  Must be called for each thread.
+ *  Has no effect if HAVE_CUDA flag is not set.
+ */
+void vosk_gpu_thread_init();
+
 #ifdef __cplusplus
 }
 #endif

java/demo/build.gradle

@@ -8,12 +8,8 @@ application {
 
 repositories {
     mavenCentral()
-    maven {
-        url 'https://alphacephei.com/maven/'
-    }
 }
 
 dependencies {
-    implementation group: 'net.java.dev.jna', name: 'jna', version: '5.7.0'
-    implementation group: 'com.alphacephei', name: 'vosk', version: '0.3.30+'
+    implementation group: 'com.alphacephei', name: 'vosk', version: '0.3.75'
 }

java/lib/build.gradle

@@ -1,18 +1,31 @@
+buildscript {
+  repositories {
+    mavenCentral()
+  }
+}
+
 plugins {
     id 'java-library'
     id 'maven-publish'
+    id 'com.vanniktech.maven.publish' version '0.18.0'
 }
 
-archivesBaseName = 'vosk'
-group = 'com.alphacephei'
-version = '0.3.30'
-
 repositories {
     mavenCentral()
 }
 
+archivesBaseName = 'vosk'
+group = 'com.alphacephei'
+version = '0.3.75'
+
+mavenPublish {
+    group = 'com.alphacephei'
+    version = version
+    sonatypeHost = 's01'
+}
+
 dependencies {
-    implementation group: 'net.java.dev.jna', name: 'jna', version: '5.7.0'
+    api group: 'net.java.dev.jna', name: 'jna', version: '5.18.1'
     testImplementation 'junit:junit:4.13'
 }
 
@@ -45,14 +58,14 @@ publishing {
             }
         }
     }
-    repositories {
-        maven {
-            url = "repo"
-        }
-    }
 }
 
 test {
     dependsOn cleanTest
     testLogging.showStandardStreams = true
 }
+
+java {
+    withSourcesJar()
+    withJavadocJar()
+}

java/lib/src/main/java/org/vosk/LibVosk.java

@@ -1,7 +1,6 @@
 package org.vosk;
 
 import com.sun.jna.Native;
-import com.sun.jna.Library;
 import com.sun.jna.Platform;
 import com.sun.jna.Pointer;
 import java.io.File;
@@ -13,8 +12,9 @@ import java.nio.file.StandardCopyOption;
 public class LibVosk {
 
     private static void unpackDll(File targetDir, String lib) throws IOException {
-        InputStream source = LibVosk.class.getResourceAsStream("/win32-x86-64/" + lib + ".dll");
-        Files.copy(source, new File(targetDir, lib + ".dll").toPath(), StandardCopyOption.REPLACE_EXISTING);
+        try (InputStream source = LibVosk.class.getResourceAsStream("/win32-x86-64/" + lib + ".dll")) {
+            Files.copy(source, new File(targetDir, lib + ".dll").toPath(), StandardCopyOption.REPLACE_EXISTING);
+        }
     }
 
     static {
@@ -23,7 +23,7 @@ public class LibVosk {
             // We have to unpack dependencies
             try {
                 // To get a tmp folder we unpack small library and mark it for deletion
-                File tmpFile = Native.extractFromResourcePath("/win32-x86-64/empty");
+                File tmpFile = Native.extractFromResourcePath("/win32-x86-64/empty", LibVosk.class.getClassLoader());
                 File tmpDir = tmpFile.getParentFile();
                 new File(tmpDir, tmpFile.getName() + ".x").createNewFile();
 
@@ -62,6 +62,8 @@ public class LibVosk {
 
     public static native void vosk_recognizer_set_words(Pointer recognizer, boolean words);
 
+    public static native void vosk_recognizer_set_partial_words(Pointer recognizer, boolean partial_words);
+
     public static native void vosk_recognizer_set_spk_model(Pointer recognizer, Pointer spk_model);
 
     public static native boolean vosk_recognizer_accept_waveform(Pointer recognizer, byte[] data, int len);
@@ -76,10 +78,30 @@ public class LibVosk {
 
     public static native String vosk_recognizer_partial_result(Pointer recognizer);
 
+    public static native void vosk_recognizer_set_grm(Pointer recognizer, String grammar);
+
     public static native void vosk_recognizer_reset(Pointer recognizer);
 
+    public static native void vosk_recognizer_set_endpointer_mode(Pointer recognizer, int mode);
+
+    public static native void vosk_recognizer_set_endpointer_delays(Pointer recognizer, float t_start_max, float t_end, float t_max);
+
     public static native void vosk_recognizer_free(Pointer recognizer);
 
+    public static native Pointer vosk_text_processor_new(String verbalizer, String tagger);
+
+    public static native void vosk_text_processor_free(Pointer processor);
+
+    public static native String vosk_text_processor_itn(Pointer processor, String input);
+
+    /**
+     * Set log level for Kaldi messages.
+     *
+     *  @param loglevel the level
+     *     0 - default value to print info and error messages but no debug
+     *     less than 0 - don't print info messages
+     *     greater than 0 - more verbose mode
+     */
     public static void setLogLevel(LogLevel loglevel) {
         vosk_set_log_level(loglevel.getValue());
     }

java/lib/src/main/java/org/vosk/Model.java

@@ -1,13 +1,18 @@
 package org.vosk;
 
+import java.io.IOException;
 import com.sun.jna.PointerType;
 
 public class Model extends PointerType implements AutoCloseable {
     public Model() {
     }
 
-    public Model(String path) {
+    public Model(String path) throws IOException {
         super(LibVosk.vosk_model_new(path));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a model");
+        }
     }
 
     @Override

java/lib/src/main/java/org/vosk/Recognizer.java

@@ -1,32 +1,163 @@
 package org.vosk;
 
 import com.sun.jna.PointerType;
+import java.io.IOException;
 
 public class Recognizer extends PointerType implements AutoCloseable {
-    public Recognizer(Model model, float sampleRate) {
+    /**
+     * Creates the recognizer object.
+     *
+     * The recognizers process the speech and return text using shared model data
+     * @param model       VoskModel containing static data for recognizer. Model can be
+     *                    shared across recognizers, even running in different threads.
+     * @param sampleRate The sample rate of the audio you are going to feed into the recognizer.
+     *                    Make sure this rate matches the audio content, it is a common
+     *                    issue causing accuracy problems.
+     * @throws IOException if the recognizer could not be created
+     */
+    public Recognizer(Model model, float sampleRate) throws IOException {
         super(LibVosk.vosk_recognizer_new(model, sampleRate));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a recognizer");
+        }
     }
 
-    public Recognizer(Model model, float sampleRate, SpeakerModel spkModel) {
+    /**
+     * Creates the recognizer object with speaker recognition.
+     *
+     *  With the speaker recognition mode the recognizer not just recognize
+     *  text but also return speaker vectors one can use for speaker identification
+     *
+     *  @param model       VoskModel containing static data for recognizer. Model can be
+     *                     shared across recognizers, even running in different threads.
+     *  @param sampleRate The sample rate of the audio you are going to feed into the recognizer.
+     *                     Make sure this rate matches the audio content, it is a common
+     *                     issue causing accuracy problems.
+     *  @param spkModel speaker model for speaker identification
+     *  @throws IOException if the recognizer could not be created
+     */
+    public Recognizer(Model model, float sampleRate, SpeakerModel spkModel) throws IOException {
         super(LibVosk.vosk_recognizer_new_spk(model.getPointer(), sampleRate, spkModel.getPointer()));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a recognizer");
+        }
     }
 
-    public Recognizer(Model model, float sampleRate, String grammar) {
+    /**
+     * Creates the recognizer object with the phrase list.
+     *
+     *  Sometimes when you want to improve recognition accuracy and when you don't need
+     *  to recognize large vocabulary you can specify a list of phrases to recognize. This
+     *  will improve recognizer speed and accuracy but might return [unk] if user said
+     *  something different.
+     *
+     *  Only recognizers with lookahead models support this type of quick configuration.
+     *  Precompiled HCLG graph models are not supported.
+     *
+     *  @param model       VoskModel containing static data for recognizer. Model can be
+     *                     shared across recognizers, even running in different threads.
+     *  @param sampleRate The sample rate of the audio you are going to feed into the recognizer.
+     *                     Make sure this rate matches the audio content, it is a common
+     *                     issue causing accuracy problems.
+     *  @param grammar The string with the list of phrases to recognize as JSON array of strings,
+     *                 for example "["one two three four five", "[unk]"]".
+     *  @throws IOException if the recognizer could not be created
+     */
+    public Recognizer(Model model, float sampleRate, String grammar) throws IOException {
         super(LibVosk.vosk_recognizer_new_grm(model.getPointer(), sampleRate, grammar));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a recognizer");
+        }
     }
 
+    /**
+     * Configures recognizer to output n-best results.
+     *
+     * <pre>
+     *   {
+     *      "alternatives": [
+     *          { "text": "one two three four five", "confidence": 0.97 },
+     *          { "text": "one two three for five", "confidence": 0.03 },
+     *      ]
+     *   }
+     * </pre>
+     *
+     * @param maxAlternatives - maximum alternatives to return from recognition results
+     */
     public void setMaxAlternatives(int maxAlternatives) {
         LibVosk.vosk_recognizer_set_max_alternatives(this.getPointer(), maxAlternatives);
     }
 
+    /** Enables words with times in the output
+     *
+     * <pre>
+     *   "result" : [{
+     *       "conf" : 1.000000,
+     *       "end" : 1.110000,
+     *       "start" : 0.870000,
+     *       "word" : "what"
+     *     }, {
+     *       "conf" : 1.000000,
+     *       "end" : 1.530000,
+     *       "start" : 1.110000,
+     *       "word" : "zero"
+     *     }, {
+     *       "conf" : 1.000000,
+     *       "end" : 1.950000,
+     *       "start" : 1.530000,
+     *       "word" : "zero"
+     *     }, {
+     *       "conf" : 1.000000,
+     *       "end" : 2.340000,
+     *       "start" : 1.950000,
+     *       "word" : "zero"
+     *     }, {
+     *       "conf" : 1.000000,
+     *       "end" : 2.610000,
+     *       "start" : 2.340000,
+     *       "word" : "one"
+     *     }],
+     * </pre>
+     *
+     * @param words - boolean value
+     */
     public void setWords(boolean words) {
         LibVosk.vosk_recognizer_set_words(this.getPointer(), words);
     }
 
+    /**
+     * Like above return words and confidences in partial results.
+     *
+     * @param partial_words - boolean value
+     */
+    public void setPartialWords(boolean partial_words) {
+        LibVosk.vosk_recognizer_set_partial_words(this.getPointer(), partial_words);
+    }
+
+    /**
+     * Adds speaker model to already initialized recognizer.
+     *
+     * Can add speaker recognition model to already created recognizer.
+     * Helps to initialize speaker recognition for grammar-based recognizer.
+     *
+     * @param spkModel Speaker recognition model
+     */
     public void setSpeakerModel(SpeakerModel spkModel) {
         LibVosk.vosk_recognizer_set_spk_model(this.getPointer(), spkModel.getPointer());
     }
 
+    /**
+     * Accept and process new chunk of voice data.
+     *
+     *  @param data - audio data in PCM 16-bit mono format
+     *  @param len - length of the audio data
+     *  @return 1 if silence is occurred and you can retrieve a new utterance with result method
+     *           0 if decoding continues
+     *           -1 if exception occurred
+     */
     public boolean acceptWaveForm(byte[] data, int len) {
         return LibVosk.vosk_recognizer_accept_waveform(this.getPointer(), data, len);
     }
@@ -39,22 +170,104 @@ public class Recognizer extends PointerType implements AutoCloseable {
         return LibVosk.vosk_recognizer_accept_waveform_f(this.getPointer(), data, len);
     }
 
+    /**
+     * Returns speech recognition result
+     *
+     * @return the result in JSON format which contains decoded line, decoded
+     *          words, times in seconds and confidences. You can parse this result
+     *          with any json parser
+     *
+     * <pre>
+     *  {
+     *    "text" : "what zero zero zero one"
+     *  }
+     * </pre>
+     *
+     * If alternatives enabled it returns result with alternatives, see also #setMaxAlternatives().
+     *
+     * If word times enabled returns word time, see also #setWordTimes().
+     */
     public String getResult() {
         return LibVosk.vosk_recognizer_result(this.getPointer());
     }
 
+    /**
+     * Returns partial speech recognition.
+     *
+     * @return partial speech recognition text which is not yet finalized.
+     *          result may change as recognizer process more data.
+     *
+     * <pre>
+     * {
+     *    "partial" : "cyril one eight zero"
+     * }
+     * </pre>
+     */
     public String getPartialResult() {
         return LibVosk.vosk_recognizer_partial_result(this.getPointer());
     }
 
+    /**
+     * Returns speech recognition result. Same as result, but doesn't wait for silence.
+     *  You usually call it in the end of the stream to get final bits of audio. It
+     *  flushes the feature pipeline, so all remaining audio chunks got processed.
+     *
+     *  @return speech result in JSON format.
+     */
     public String getFinalResult() {
         return LibVosk.vosk_recognizer_final_result(this.getPointer());
     }
 
+    /**
+     * Reconfigures recognizer to use grammar.
+     *
+     * @param grammar      Set of phrases in JSON array of strings or "[]" to use default model graph.
+     * @see #Recognizer(Model, float, String)
+     */
+    public void setGrammar(String grammar) {
+        LibVosk.vosk_recognizer_set_grm(this.getPointer(), grammar);
+    }
+
+    /**
+     * Resets the recognizer.
+     * Resets current results so the recognition can continue from scratch.
+     */
     public void reset() {
         LibVosk.vosk_recognizer_reset(this.getPointer());
     }
 
+    /**
+     * Endpointer delay mode
+     */
+    public class EndpointerMode {
+        public static final int DEFAULT = 0;
+        public static final int SHORT = 1;
+        public static final int LONG = 2;
+        public static final int VERY_LONG = 3;
+    }
+
+    /**
+     * Configures endpointer mode for recognizer
+     */
+    public void setEndpointerMode(int mode) {
+        LibVosk.vosk_recognizer_set_endpointer_mode(this.getPointer(), mode);
+    }
+
+    /**
+     * Set endpointer delays
+     *
+     * @param t_start_max     timeout for stopping recognition in case of initial silence (usually around 5.0)
+     * @param t_end           timeout for stopping recognition in milliseconds after we recognized something (usually around 0.5 - 1.0)
+     * @param t_max           timeout for forcing utterance end in milliseconds (usually around 20-30)
+     **/
+    public void setEndpointerDelays(float t_start_max, float t_end, float t_max) {
+        LibVosk.vosk_recognizer_set_endpointer_delays(this.getPointer(), t_start_max, t_end, t_max);
+    }
+
+    /**
+     * Releases recognizer object.
+     * Underlying model is also unreferenced and if needed, released.
+     */
     @Override
     public void close() {
         LibVosk.vosk_recognizer_free(this.getPointer());

java/lib/src/main/java/org/vosk/SpeakerModel.java

@@ -1,13 +1,36 @@
 package org.vosk;
 
 import com.sun.jna.PointerType;
+import java.io.IOException;
 
+/**
+ * Helps to initialize speaker recognition for grammar-based recognizer.
+ */
 public class SpeakerModel extends PointerType implements AutoCloseable {
     public SpeakerModel() {
     }
 
-    public SpeakerModel(String path) {
+    /**
+     * Loads speaker model data from the file.
+     *
+     * The path must contain:
+     * - a config file: mfcc.conf
+     * - kaldi nnet: final.ext.raw
+     * - mean.vec
+     * - transform.mat
+     *
+     * @param path the path of the model on the filesystem
+     * @throws IOException if the model could not be created
+     *
+     * @see <a href="http://kaldi-asr.org/doc/structkaldi_1_1MfccOptions.html">Kaldi MfccOptions</a>
+     * @see <a href="http://kaldi-asr.org/doc/classkaldi_1_1nnet3_1_1Nnet.html">Kaldi Nnet</a>
+     */
+    public SpeakerModel(String path) throws IOException {
         super(LibVosk.vosk_spk_model_new(path));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a speaker model");
+        }
     }
 
     @Override

java/lib/src/test/java/org/vosk/test/DecoderTest.java

@@ -15,8 +15,10 @@ import javax.sound.sampled.UnsupportedAudioFileException;
 
 import org.vosk.LogLevel;
 import org.vosk.Recognizer;
+import org.vosk.Recognizer.EndpointerMode;
 import org.vosk.LibVosk;
 import org.vosk.Model;
+import org.vosk.TextProcessor;
 
 public class DecoderTest {
 
@@ -30,6 +32,7 @@ public class DecoderTest {
 
             recognizer.setMaxAlternatives(10);
             recognizer.setWords(true);
+            recognizer.setPartialWords(true);
 
             int nbytes;
             byte[] b = new byte[4096];
@@ -93,4 +96,25 @@ public class DecoderTest {
         }
         Assert.assertTrue(true);
     }
+
+    @Test
+    public void decoderEndpointerDelays() throws IOException, UnsupportedAudioFileException {
+        try (Model model = new Model("model");
+            Recognizer recognizer = new Recognizer(model, 16000)) {
+            recognizer.setEndpointerMode(EndpointerMode.VERY_LONG);
+            recognizer.setEndpointerDelays(5.0f, 3.0f, 50.0f);
+	}
+        Assert.assertTrue(true);
+    }
+
+    @Test(expected = IOException.class)
+    public void decoderTestException() throws IOException {
+        Model model = new Model("model_missing");
+    }
+
+    @Test
+    public void testItn() throws IOException {
+        TextProcessor p = new TextProcessor("model/itn/en_itn_tagger.fst", "model/itn/en_itn_verbalizer.fst");
+        System.out.println(p.itn("as easy as one two three"));
+    }
 }

kotlin/.gitignore

@@ -0,0 +1,3 @@
+.gradle/
+.idea/
+build/

kotlin/AUTHORS

@@ -0,0 +1 @@
+Doomsdayrs doomsdayrs@gmail.com

kotlin/MAINTENANCE.md

@@ -0,0 +1,8 @@
+# Maintenance
+
+To maintain this module, please ensure the following.
+1. Kotlin version is kept up to date.
+This will be found in the plugins block of the [build.gradle.kts](./build.gradle.kts).
+Ensure both multiplatform and serialization have the same value.
+2. Ensure dependencies are up to date
+3. Ensure that the android min & target sdks are up to date.

kotlin/README.md

@@ -0,0 +1,63 @@
+# vosk-api-kotlin
+
+The vosk-api wrapped using Kotlin Multiplatform.
+
+## Usage
+
+The following are ways to use this wrapper.
+
+### JVM
+
+For Java & Android targets.
+
+```kotlin
+dependencoes {
+    val voskVersion = "0.4.0-alpha0"
+
+    // Generic
+    implementation("com.alphacephei:vosk-api-kotlin:$voskVersion")
+
+    // Android
+    implementation("com.alphacephei:vosk-api-kotlin-android:$voskVersion")
+}
+```
+
+## Building
+
+To build this project, follow the following steps.
+
+1. Install `libvosk`.
+This can be done from [source][source install] (parent monorepo) 
+or [downloaded][download].
+2. Download a [Vosk model](https://alphacephei.com/vosk/models) to use.
+  - It is suggested to use a small model to speed up tests.
+3. Once both are downloaded and placed into a proper location (hopefully following UNIX specification).
+Set the following environment variables:
+  - `VOSK_MODEL` to the path of the model.
+  - `VOSK_PATH` to the path of `libvosk`.
+These are used by the various tests to operate.
+4. Now that the required steps are complete, one can run `./gradlew build`.
+
+## Kotlin/Native
+
+Currently, the native target is disabled due to a lack of vosk-api packaging on platforms.
+Further worsened by the fact that installing the vosk-api on Linux systems is a chore.
+
+First, either install from [source][source install]
+or [download][download] and install into the proper unix directories as expected in
+[libvosk.def](./src/nativeInterop/cinterop/libvosk.def).
+
+To enable development for Kotlin/Native, do either for the following.
+- Add `NATIVE_EXPERIMENT=true` to your environment.
+- Go into [build.gradle.kts](./build.gradle.kts) & find `enableNative` & set the right side to true.
+
+Afterwards, when syncing the project, the native source sets will become available to work on.
+It is suggested to run `cinteropLibvoskNative` to generate the Kotlin C bindings to work with.
+
+## Future
+
+- Possibly target Kotlin/JS
+- Possibly target Kotlin/Objective-C (?)
+
+[source install]: https://alphacephei.com/vosk/install
+[download]: https://github.com/alphacep/vosk-api/releases/latest

kotlin/build.gradle.kts

@@ -0,0 +1,204 @@
+import org.jetbrains.dokka.gradle.DokkaTask
+import org.jetbrains.kotlin.gradle.ExperimentalKotlinGradlePluginApi
+import org.jetbrains.kotlin.gradle.dsl.JvmTarget
+
+/*
+ * Copyright 2020 Alpha Cephei Inc. & Doomsdayrs
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+plugins {
+	kotlin("multiplatform") version "2.0.0"
+	id("com.android.library")
+	`maven-publish`
+	id("org.jetbrains.dokka") version "1.9.20"
+	kotlin("plugin.serialization") version "2.0.0"
+}
+
+group = "com.alphacephei"
+version = "0.3.75"
+
+repositories {
+	google()
+	mavenCentral()
+}
+
+val dokkaOutputDir = "$buildDir/dokka"
+
+tasks.getByName<DokkaTask>("dokkaHtml") {
+	outputDirectory.set(file(dokkaOutputDir))
+}
+
+val deleteDokkaOutputDir by tasks.register<Delete>("deleteDokkaOutputDirectory") {
+	delete(dokkaOutputDir)
+}
+
+val javadocJar = tasks.register<Jar>("javadocJar") {
+	dependsOn(deleteDokkaOutputDir, tasks.dokkaHtml)
+	archiveClassifier.set("javadoc")
+	from(dokkaOutputDir)
+}
+
+fun org.jetbrains.kotlin.gradle.dsl.KotlinMultiplatformExtension.native(
+	configure: org.jetbrains.kotlin.gradle.plugin.mpp.KotlinNativeTargetWithHostTests.() -> Unit = {}
+) {
+	when {
+		org.jetbrains.kotlin.konan.target.HostManager.hostIsMingw -> mingwX64("native")
+		org.jetbrains.kotlin.konan.target.HostManager.hostIsLinux -> linuxX64("native")
+		org.jetbrains.kotlin.konan.target.HostManager.hostIsMac -> if (org.jetbrains.kotlin.konan.target.HostManager.hostArch() == "arm64") {
+			macosArm64("native")
+		} else {
+			macosX64("native")
+		}
+
+		else -> error("Unsupported Host OS: ${org.jetbrains.kotlin.konan.target.HostManager.hostOs()}")
+	}.apply(configure)
+}
+
+kotlin {
+	jvm {
+		@OptIn(ExperimentalKotlinGradlePluginApi::class)
+		compilerOptions {
+			jvmTarget.set(JvmTarget.JVM_17)
+		}
+
+		testRuns["test"].executionTask.configure {
+			useJUnitPlatform()
+			environment("MODEL", "VOSK_MODEL")
+			//environment("MODEL", "/home/doomsdayrs/Downloads/vosk-model-small-en-us-0.15/")
+			environment("LIBRARY", "VOSK_PATH")
+			//environment("LIBRARY", "/usr/local/lib64/libvosk/libvosk.so")
+			environment("AUDIO", "$projectDir/../python/example/test.wav")
+		}
+	}
+
+	androidTarget {
+		publishAllLibraryVariants()
+	}
+
+	/**
+	 * If native target should be enabled or not.
+	 *
+	 * Currently disabled as there is no proper packaging distribution currently.
+	 */
+	@Suppress("SimplifyBooleanWithConstants") // Ignore, the false is for overrides
+	val enableNative = System.getenv("NATIVE_EXPERIMENT") == "true" || false
+
+	if (enableNative)
+		native {
+			val main by compilations.getting
+			val libvosk by main.cinterops.creating
+
+			binaries {
+				sharedLib()
+			}
+		}
+
+
+	@OptIn(ExperimentalKotlinGradlePluginApi::class)
+	applyDefaultHierarchyTemplate {
+		withJvm()
+		withAndroidTarget()
+
+		if (enableNative)
+			withNative()
+	}
+
+	publishing {
+		publications {
+			withType<MavenPublication> {
+				artifact(javadocJar)
+				pom {
+					url.set("http://www.alphacephei.com.com/vosk/")
+					licenses {
+						license {
+							name.set("The Apache License, Version 2.0")
+							url.set("http://www.apache.org/licenses/LICENSE-2.0.txt")
+						}
+					}
+					developers {
+						developer {
+							id.set("com.alphacephei")
+							name.set("Alpha Cephei Inc")
+							email.set("contact@alphacephei.com")
+						}
+					}
+					scm {
+						connection.set("scm:git:git://github.com/alphacep/vosk-api.git")
+						url.set("https://github.com/alphacep/vosk-api/")
+					}
+				}
+			}
+		}
+	}
+
+	val jna_version = "5.14.0"
+	val coroutines_version = "1.7.3"
+
+	sourceSets {
+		val commonMain by getting {
+			dependencies {
+				api("org.jetbrains.kotlinx:kotlinx-serialization-json:1.7.0")
+				api("org.jetbrains.kotlinx:kotlinx-coroutines-core:$coroutines_version")
+			}
+		}
+		val commonTest by getting {
+			dependencies {
+				implementation(kotlin("test"))
+				implementation("org.jetbrains.kotlinx:kotlinx-coroutines-test:$coroutines_version")
+			}
+		}
+		val jvmMain by getting {
+			dependencies {
+				api("net.java.dev.jna:jna:$jna_version")
+			}
+		}
+		val jvmTest by getting
+		if (enableNative) {
+			val nativeMain by getting
+		}
+		val androidMain by getting {
+			dependsOn(jvmMain)
+			dependencies {
+				api("net.java.dev.jna:jna:$jna_version@aar")
+			}
+		}
+		val androidUnitTest by getting {
+			dependencies {
+				implementation("junit:junit:4.13.2")
+			}
+		}
+	}
+}
+
+android {
+	namespace = "com.alphacephei.library"
+	compileSdk = 34
+	sourceSets["main"].manifest.srcFile("src/androidMain/AndroidManifest.xml")
+	defaultConfig {
+		minSdk = 24
+		targetSdk = 34
+	}
+	compileOptions {
+		sourceCompatibility = JavaVersion.VERSION_17
+		targetCompatibility = JavaVersion.VERSION_17
+	}
+	publishing {
+		multipleVariants {
+			withSourcesJar()
+			withJavadocJar()
+			allVariants()
+		}
+	}
+}

kotlin/settings.gradle.kts

@@ -0,0 +1,31 @@
+/*
+ * Copyright 2020 Alpha Cephei Inc. & Doomsdayrs
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+pluginManagement {
+	repositories {
+		google()
+		gradlePluginPortal()
+		mavenCentral()
+	}
+	resolutionStrategy {
+		eachPlugin {
+			if (requested.id.namespace == "com.android") {
+				useModule("com.android.tools.build:gradle:8.3.0")
+			}
+		}
+	}
+}
+rootProject.name = "vosk-api-kotlin"

kotlin/src/androidMain/AndroidManifest.xml

@@ -0,0 +1,6 @@
+<?xml version="1.0" encoding="utf-8"?>
+<manifest xmlns:android="http://schemas.android.com/apk/res/android"
+    package="com.alphacephei.library">
+
+    <uses-permission android:name="android.permission.RECORD_AUDIO" />
+</manifest>

kotlin/src/androidMain/kotlin/org/vosk/android/RecognitionListener.kt

@@ -0,0 +1,46 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.vosk.android
+
+/**
+ * Interface to receive recognition results
+ */
+interface RecognitionListener {
+	/**
+	 * Called when partial recognition result is available.
+	 */
+	fun onPartialResult(hypothesis: String)
+
+	/**
+	 * Called after silence occured.
+	 */
+	fun onResult(hypothesis: String)
+
+	/**
+	 * Called after stream end.
+	 */
+	fun onFinalResult(hypothesis: String)
+
+	/**
+	 * Called when an error occurs.
+	 */
+	fun onError(exception: Exception)
+
+	/**
+	 * Called after timeout expired
+	 */
+	fun onTimeout()
+}

kotlin/src/androidMain/kotlin/org/vosk/android/SpeechService.kt

@@ -0,0 +1,241 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.vosk.android
+
+import android.annotation.SuppressLint
+import android.media.AudioFormat
+import android.media.AudioRecord
+import android.media.MediaRecorder.AudioSource
+import android.os.Handler
+import android.os.Looper
+import org.vosk.Recognizer
+import java.io.IOException
+import kotlin.math.roundToInt
+
+/**
+ * Service that records audio in a thread, passes it to a recognizer and emits
+ * recognition results. Recognition events are passed to a client using
+ * [RecognitionListener]
+ */
+class SpeechService @Throws(IOException::class) constructor(
+	private val recognizer: Recognizer,
+	sampleRate: Float
+) {
+	private val sampleRate: Int
+	private val bufferSize: Int
+	private val recorder: AudioRecord
+	private var recognizerThread: RecognizerThread? = null
+	private val mainHandler = Handler(Looper.getMainLooper())
+
+	/**
+	 * Creates speech service. Service holds the AudioRecord object, so you
+	 * need to call [.shutdown] in order to properly finalize it.
+	 *
+	 * @throws IOException thrown if audio recorder can not be created for some reason.
+	 */
+	init {
+		this.sampleRate = sampleRate.toInt()
+		bufferSize = (this.sampleRate * BUFFER_SIZE_SECONDS).roundToInt()
+		@SuppressLint("MissingPermission")
+		recorder = AudioRecord(
+			AudioSource.VOICE_RECOGNITION, this.sampleRate,
+			AudioFormat.CHANNEL_IN_MONO,
+			AudioFormat.ENCODING_PCM_16BIT, bufferSize * 2
+		)
+		if (recorder.state == AudioRecord.STATE_UNINITIALIZED) {
+			recorder.release()
+			throw IOException(
+				"Failed to initialize recorder. Microphone might be already in use."
+			)
+		}
+	}
+
+	/**
+	 * Starts recognition. Does nothing if recognition is active.
+	 *
+	 * @return true if recognition was actually started
+	 */
+	fun startListening(listener: RecognitionListener): Boolean {
+		if (null != recognizerThread) return false
+		recognizerThread = RecognizerThread(listener)
+		recognizerThread!!.start()
+		return true
+	}
+
+	/**
+	 * Starts recognition. After specified timeout listening stops and the
+	 * endOfSpeech signals about that. Does nothing if recognition is active.
+	 *
+	 *
+	 * timeout - timeout in milliseconds to listen.
+	 *
+	 * @return true if recognition was actually started
+	 */
+	fun startListening(listener: RecognitionListener, timeout: Int): Boolean {
+		if (null != recognizerThread) return false
+		recognizerThread = RecognizerThread(listener, timeout)
+		recognizerThread!!.start()
+		return true
+	}
+
+	private fun stopRecognizerThread(): Boolean {
+		if (null == recognizerThread) return false
+		try {
+			recognizerThread!!.interrupt()
+			recognizerThread!!.join()
+		} catch (e: InterruptedException) {
+			// Restore the interrupted status.
+			Thread.currentThread().interrupt()
+		}
+		recognizerThread = null
+		return true
+	}
+
+	/**
+	 * Stops recognition. Listener should receive final result if there is
+	 * any. Does nothing if recognition is not active.
+	 *
+	 * @return true if recognition was actually stopped
+	 */
+	fun stop(): Boolean {
+		return stopRecognizerThread()
+	}
+
+	/**
+	 * Cancel recognition. Do not post any new events, simply cancel processing.
+	 * Does nothing if recognition is not active.
+	 *
+	 * @return true if recognition was actually stopped
+	 */
+	fun cancel(): Boolean {
+		if (recognizerThread != null) {
+			recognizerThread!!.setPause(true)
+		}
+		return stopRecognizerThread()
+	}
+
+	/**
+	 * Shutdown the recognizer and release the recorder
+	 */
+	fun shutdown() {
+		recorder.release()
+	}
+
+	fun setPause(paused: Boolean) {
+		if (recognizerThread != null) {
+			recognizerThread!!.setPause(paused)
+		}
+	}
+
+	/**
+	 * Resets recognizer in a thread, starts recognition over again
+	 */
+	fun reset() {
+		if (recognizerThread != null) {
+			recognizerThread!!.reset()
+		}
+	}
+
+	private inner class RecognizerThread @JvmOverloads constructor(
+		var listener: RecognitionListener,
+		timeout: Int = Companion.NO_TIMEOUT
+	) : Thread() {
+		private var remainingSamples: Int
+		private val timeoutSamples: Int
+
+		@Volatile
+		private var paused = false
+
+		@Volatile
+		private var reset = false
+
+		init {
+			timeoutSamples = if (timeout != Companion.NO_TIMEOUT) {
+				timeout * sampleRate / 1000
+			} else {
+				Companion.NO_TIMEOUT
+			}
+			remainingSamples = timeoutSamples
+		}
+
+		/**
+		 * When we are paused, don't process audio by the recognizer and don't emit
+		 * any listener results
+		 *
+		 * @param paused the status of pause
+		 */
+		fun setPause(paused: Boolean) {
+			this.paused = paused
+		}
+
+		/**
+		 * Set reset state to signal reset of the recognizer and start over
+		 */
+		fun reset() {
+			reset = true
+		}
+
+		override fun run() {
+			recorder.startRecording()
+			if (recorder.recordingState == AudioRecord.RECORDSTATE_STOPPED) {
+				recorder.stop()
+				val ioe = IOException(
+					"Failed to start recording. Microphone might be already in use."
+				)
+				mainHandler.post { listener.onError(ioe) }
+			}
+			val buffer = ShortArray(bufferSize)
+			while (!interrupted()
+				&& (timeoutSamples == Companion.NO_TIMEOUT || remainingSamples > 0)
+			) {
+				val nread = recorder.read(buffer, 0, buffer.size)
+				if (paused) {
+					continue
+				}
+				if (reset) {
+					recognizer.reset()
+					reset = false
+				}
+				if (nread < 0) throw RuntimeException("error reading audio buffer")
+				if (recognizer.acceptWaveform(buffer)) {
+					val result = recognizer.result
+					mainHandler.post { listener.onResult(result) }
+				} else {
+					val partialResult = recognizer.partialResult
+					mainHandler.post { listener.onPartialResult(partialResult) }
+				}
+				if (timeoutSamples != NO_TIMEOUT) {
+					remainingSamples -= nread
+				}
+			}
+			recorder.stop()
+			if (!paused) {
+				// If we met timeout signal that speech ended
+				if (timeoutSamples != NO_TIMEOUT && remainingSamples <= 0) {
+					mainHandler.post { listener.onTimeout() }
+				} else {
+					val finalResult = recognizer.finalResult
+					mainHandler.post { listener.onFinalResult(finalResult) }
+				}
+			}
+		}
+	}
+
+	companion object {
+		private const val NO_TIMEOUT = -1
+		private const val BUFFER_SIZE_SECONDS = 0.2f
+	}
+}

kotlin/src/androidMain/kotlin/org/vosk/android/SpeechStreamService.kt

@@ -0,0 +1,151 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.vosk.android
+
+import android.os.Handler
+import android.os.Looper
+import org.vosk.Recognizer
+import java.io.IOException
+import java.io.InputStream
+import kotlin.math.roundToInt
+
+/**
+ * Service that recognizes stream audio in a  thread, passes it to a recognizer and emits
+ * recognition results. Recognition events are passed to a client using
+ * [RecognitionListener]
+ */
+class SpeechStreamService(
+	private val recognizer: Recognizer,
+	inputStream: InputStream,
+	sampleRate: Float
+) {
+	private val inputStream: InputStream
+	private val sampleRate: Int
+	private val bufferSize: Int
+	private var recognizerThread: Thread? = null
+	private val mainHandler = Handler(Looper.getMainLooper())
+
+	/**
+	 * Creates speech service.
+	 */
+	init {
+		this.sampleRate = sampleRate.toInt()
+		this.inputStream = inputStream
+		bufferSize = (this.sampleRate * BUFFER_SIZE_SECONDS * 2).roundToInt()
+	}
+
+	/**
+	 * Starts recognition. Does nothing if recognition is active.
+	 *
+	 * @return true if recognition was actually started
+	 */
+	fun start(listener: RecognitionListener): Boolean {
+		if (null != recognizerThread) return false
+		recognizerThread = RecognizerThread(listener)
+		recognizerThread!!.start()
+		return true
+	}
+
+	/**
+	 * Starts recognition. After specified timeout listening stops and the
+	 * endOfSpeech signals about that. Does nothing if recognition is active.
+	 *
+	 *
+	 * timeout - timeout in milliseconds to listen.
+	 *
+	 * @return true if recognition was actually started
+	 */
+	fun start(listener: RecognitionListener, timeout: Int): Boolean {
+		if (null != recognizerThread) return false
+		recognizerThread = RecognizerThread(listener, timeout)
+		recognizerThread!!.start()
+		return true
+	}
+
+	/**
+	 * Stops recognition. All listeners should receive final result if there is
+	 * any. Does nothing if recognition is not active.
+	 *
+	 * @return true if recognition was actually stopped
+	 */
+	fun stop(): Boolean {
+		if (null == recognizerThread) return false
+		try {
+			recognizerThread!!.interrupt()
+			recognizerThread!!.join()
+		} catch (e: InterruptedException) {
+			// Restore the interrupted status.
+			Thread.currentThread().interrupt()
+		}
+		recognizerThread = null
+		return true
+	}
+
+	private inner class RecognizerThread @JvmOverloads constructor(
+		var listener: RecognitionListener,
+		timeout: Int = Companion.NO_TIMEOUT
+	) : Thread() {
+		private var remainingSamples: Int
+		private val timeoutSamples: Int
+
+		init {
+			if (timeout != Companion.NO_TIMEOUT) timeoutSamples =
+				timeout * sampleRate / 1000 else timeoutSamples = Companion.NO_TIMEOUT
+			remainingSamples = timeoutSamples
+		}
+
+		override fun run() {
+			val buffer = ByteArray(bufferSize)
+			while (!interrupted()
+				&& (timeoutSamples == Companion.NO_TIMEOUT || remainingSamples > 0)
+			) {
+				try {
+					val nread = inputStream.read(buffer, 0, buffer.size)
+					if (nread < 0) {
+						break
+					} else {
+						val isSilence: Boolean = recognizer.acceptWaveform(buffer)
+						if (isSilence) {
+							val result = recognizer.result
+							mainHandler.post { listener.onResult(result) }
+						} else {
+							val partialResult = recognizer.partialResult
+							mainHandler.post { listener.onPartialResult(partialResult) }
+						}
+					}
+					if (timeoutSamples != NO_TIMEOUT) {
+						remainingSamples -= nread
+					}
+				} catch (e: IOException) {
+					mainHandler.post { listener.onError(e) }
+				}
+			}
+
+			// If we met timeout signal that speech ended
+			if (timeoutSamples != NO_TIMEOUT && remainingSamples <= 0) {
+				mainHandler.post { listener.onTimeout() }
+			} else {
+				val finalResult = recognizer.finalResult
+				mainHandler.post { listener.onFinalResult(finalResult) }
+			}
+		}
+	}
+
+	companion object {
+		private const val NO_TIMEOUT = -1
+		private const val BUFFER_SIZE_SECONDS = 0.2f
+	}
+}

kotlin/src/androidMain/kotlin/org/vosk/android/StorageService.kt

@@ -0,0 +1,138 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.vosk.android
+
+import android.content.Context
+import android.content.res.AssetManager
+import android.os.Environment
+import android.os.Handler
+import android.os.Looper
+import android.util.Log
+import org.vosk.Model
+import java.io.*
+import java.util.concurrent.Executor
+import java.util.concurrent.Executors
+import java.util.function.Consumer
+
+/**
+ * Provides utility methods to sync model files to external storage to allow
+ * C++ code access them. Relies on file named "uuid" to track updates.
+ */
+object StorageService {
+	private val TAG = StorageService::class.simpleName
+
+	@JvmStatic
+	fun unpack(
+		context: Context,
+		sourcePath: String,
+		targetPath: String,
+		completeCallback: Consumer<Model>,
+		errorCallback: Consumer<IOException>
+	) {
+		val executor: Executor =
+			Executors.newSingleThreadExecutor() // change according to your requirements
+		val handler = Handler(Looper.getMainLooper())
+		executor.execute {
+			try {
+				val outputPath = sync(context, sourcePath, targetPath)
+				val model = Model(outputPath)
+				handler.post { completeCallback.accept(model) }
+			} catch (e: IOException) {
+				handler.post { errorCallback.accept(e) }
+			}
+		}
+	}
+
+	@JvmStatic
+	@Throws(IOException::class)
+	fun sync(context: Context, sourcePath: String, targetPath: String): String {
+		val assetManager = context.assets
+		val externalFilesDir = context.getExternalFilesDir(null)
+			?: throw IOException(
+				"cannot get external files dir, "
+						+ "external storage state is " + Environment.getExternalStorageState()
+			)
+		val targetDir = File(externalFilesDir, targetPath)
+		val resultPath = File(targetDir, sourcePath).absolutePath
+		val sourceUUID = readLine(assetManager.open("$sourcePath/uuid"))
+		try {
+			val targetUUID = readLine(FileInputStream(File(targetDir, "$sourcePath/uuid")))
+			if (targetUUID == sourceUUID) return resultPath
+		} catch (e: FileNotFoundException) {
+			// ignore
+		}
+		deleteContents(targetDir)
+		copyAssets(assetManager, sourcePath, targetDir)
+
+		// Copy uuid
+		copyFile(assetManager, "$sourcePath/uuid", targetDir)
+		return resultPath
+	}
+
+	@Throws(IOException::class)
+	private fun readLine(inputStream: InputStream): String {
+		return BufferedReader(InputStreamReader(inputStream)).use { it.readLine() }
+	}
+
+	private fun deleteContents(dir: File): Boolean {
+		val files = dir.listFiles()
+		var success = true
+		if (files != null) {
+			for (file in files) {
+				if (file.isDirectory) {
+					success = success and deleteContents(file)
+				}
+				if (!file.delete()) {
+					success = false
+				}
+			}
+		}
+		return success
+	}
+
+	@Throws(IOException::class)
+	private fun copyAssets(assetManager: AssetManager, path: String, outPath: File) {
+		val assets = assetManager.list(path) ?: return
+		if (assets.isEmpty()) {
+			if (!path.endsWith("uuid")) copyFile(assetManager, path, outPath)
+		} else {
+			val dir = File(outPath, path)
+			if (!dir.exists()) {
+				Log.v(TAG, "Making directory " + dir.absolutePath)
+				if (!dir.mkdirs()) {
+					Log.v(TAG, "Failed to create directory " + dir.absolutePath)
+				}
+			}
+			for (asset in assets) {
+				copyAssets(assetManager, "$path/$asset", outPath)
+			}
+		}
+	}
+
+	@Throws(IOException::class)
+	private fun copyFile(assetManager: AssetManager, fileName: String, outPath: File) {
+		Log.v(TAG, "Copy $fileName to $outPath")
+		assetManager.open(fileName).use { inputStream ->
+			FileOutputStream("$outPath/$fileName").use { out ->
+				val buffer = ByteArray(4000)
+				var read: Int
+				while (inputStream.read(buffer).also { read = it } != -1) {
+					out.write(buffer, 0, read)
+				}
+			}
+		}
+	}
+}

kotlin/src/commonMain/kotlin/org/vosk/AcceptWaveformException.kt

@@ -0,0 +1,22 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+/**
+ * Thrown when a [Recognizer] cannot accept a given waveform
+ */
+class AcceptWaveformException(data: Any) : Exception("Could not accept waveform: $data")

kotlin/src/commonMain/kotlin/org/vosk/BatchModel.kt

@@ -0,0 +1,38 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import org.vosk.exception.ModelException
+
+/**
+ * Batch model object
+ *
+ * @since 26 / 12 / 2022
+ * @constructor Creates the batch recognizer object
+ */
+expect class BatchModel @Throws(ModelException::class) constructor(path: String) : Freeable {
+
+	/**
+	 *  Releases batch model object
+	 */
+	override fun free()
+
+	/**
+	 * Wait for the processing
+	 */
+	fun await()
+}

kotlin/src/commonMain/kotlin/org/vosk/BatchRecognizer.kt

@@ -0,0 +1,62 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+/**
+ * Batch recognizer object
+ *
+ * @since 26 / 12 / 2022
+ * @constructor Creates batch recognizer object
+ */
+expect class BatchRecognizer constructor(model: BatchModel, sampleRate: Float) : Freeable {
+
+	/**
+	 * Releases batch recognizer object
+	 */
+	override fun free()
+
+	/**
+	 * Accept batch voice data
+	 */
+	fun acceptWaveform(data: ByteArray)
+
+	/**
+	 * Set NLSML output
+	 * @param nlsml - boolean value
+	 */
+	fun setNLSML(nlsml: Boolean)
+
+	/**
+	 *  Closes the stream
+	 */
+	fun finishStream()
+
+	/**
+	 * Return results
+	 */
+	val frontResult: String
+
+	/**
+	 * Release and free first retrieved result
+	 */
+	fun pop()
+
+	/**
+	 * Get amount of pending chunks for more intelligent waiting
+	 */
+	val pendingChunks: Int
+}

kotlin/src/commonMain/kotlin/org/vosk/EndPointerMode.kt

@@ -0,0 +1,27 @@
+/*
+ * Copyright 2024 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+/**
+ * VoskEpMode
+ */
+enum class EndPointerMode {
+	ANSWER_DEFAULT,
+	ANSWER_SHORT,
+	ANSWER_LONG,
+	ANSWER_VERY_LONG
+}

kotlin/src/commonMain/kotlin/org/vosk/Freeable.kt

@@ -0,0 +1,31 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+/**
+ * Denotes an object that must be freed afterwards.
+ *
+ * On JVM, This is done via AutoClosable.
+ */
+@Suppress("SpellCheckingInspection")
+interface Freeable {
+
+	/**
+	 * Dereference the related object
+	 */
+	fun free()
+}

kotlin/src/commonMain/kotlin/org/vosk/LogLevel.kt

@@ -0,0 +1,40 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+/**
+ * Log level for Kaldi messages.
+ *
+ * @since 26 / 12 / 2022
+ */
+enum class LogLevel(val value: Int) {
+
+	/**
+	 * Don't print info messages
+	 */
+	WARNINGS(-1),
+
+	/**
+	 * Default value to print info and error messages but no debug
+	 */
+	INFO(0),
+
+	/**
+	 * More verbose mode
+	 */
+	DEBUG(1);
+}

kotlin/src/commonMain/kotlin/org/vosk/Model.kt

@@ -0,0 +1,52 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import org.vosk.exception.IOException
+
+
+/**
+ * Model stores all the data required for recognition
+ *
+ * It contains static data and can be shared across processing
+ * threads.
+ *
+ * @since 26 / 12 / 2022
+ * @constructor Loads model data from the file and returns the model object
+ * @param path the path of the model on the filesystem
+ * @throws IOException if the path provided is invalid
+ */
+expect class Model @Throws(IOException::class) constructor(path: String) : Freeable {
+
+	/**
+	 * Check if a word can be recognized by the model
+	 * @param word: the word
+	 * @returns the word symbol if @param word exists inside the model
+	 * or -1 otherwise.
+	 * Reminding that word symbol 0 is for <epsilon>
+	 */
+	fun findWord(word: String): Int
+
+	/**
+	 * Releases the model memory
+	 *
+	 *  The model object is reference-counted so if some recognizer
+	 *  depends on this model, model might still stay alive. When
+	 *  last recognizer is released, model will be released too.
+	 */
+	override fun free()
+}

kotlin/src/commonMain/kotlin/org/vosk/Recognizer.kt

@@ -0,0 +1,273 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import org.vosk.exception.RecognizerException
+
+/**
+ * Recognizer object is the main object which processes data.
+ *
+ * Each recognizer usually runs in own thread and takes audio as input.
+ * Once audio is processed recognizer returns JSON object as a string
+ * which represent decoded information - words, confidences, times, n-best lists,
+ * speaker information and so on
+ *
+ * @since 26 / 12 / 2022
+ */
+expect class Recognizer : Freeable {
+	/**
+	 * Creates the recognizer object
+	 *
+	 *  The recognizers process the speech and return text using shared model data
+	 *  @param model       VoskModel containing static data for recognizer. Model can be
+	 *                     shared across recognizers, even running in different threads.
+	 *  @param sampleRate The sample rate of the audio you going to feed into the recognizer.
+	 *                     Make sure this rate matches the audio content, it is a common
+	 *                     issue causing accuracy problems.
+	 * @throws RecognizerException if a problem occurred
+	 */
+	@Throws(RecognizerException::class)
+	constructor(model: Model, sampleRate: Float)
+
+	/**
+	 * Creates the recognizer object with speaker recognition
+	 *
+	 *  With the speaker recognition mode the recognizer not just recognize
+	 *  text but also return speaker vectors one can use for speaker identification
+	 *
+	 *  @param model       VoskModel containing static data for recognizer. Model can be
+	 *                     shared across recognizers, even running in different threads.
+	 *  @param sampleRate The sample rate of the audio you going to feed into the recognizer.
+	 *                     Make sure this rate matches the audio content, it is a common
+	 *                     issue causing accuracy problems.
+	 *  @param speakerModel speaker model for speaker identification
+	 * @throws RecognizerException if a problem occurred
+	 */
+	@Throws(RecognizerException::class)
+	constructor(model: Model, sampleRate: Float, speakerModel: SpeakerModel)
+
+	/**
+	 * Creates the recognizer object with the phrase list
+	 *
+	 *  Sometimes when you want to improve recognition accuracy and when you don't need
+	 *  to recognize large vocabulary you can specify a list of phrases to recognize. This
+	 *  will improve recognizer speed and accuracy but might return [unk] if user said
+	 *  something different.
+	 *
+	 *  Only recognizers with lookahead models support this type of quick configuration.
+	 *  Precompiled HCLG graph models are not supported.
+	 *
+	 *  @param model       VoskModel containing static data for recognizer. Model can be
+	 *                     shared across recognizers, even running in different threads.
+	 *  @param sampleRate The valuesample rate of the audio you going to feed into the recognizer.
+	 *                     Make sure this rate matches the audio content, it is a common
+	 *                     issue causing accuracy problems.
+	 *  @param grammar The string with the list of phrases to recognize as JSON array of strings,
+	 *                 for example "["one two three four five", "[unk]"]".
+	 *
+	 * @throws RecognizerException if a problem occurred
+	 */
+	@Throws(RecognizerException::class)
+	constructor(model: Model, sampleRate: Float, grammar: String)
+
+	/**
+	 * Adds speaker model to already initialized recognizer
+	 *
+	 * Can add speaker recognition model to already created recognizer. Helps to initialize
+	 * speaker recognition for grammar-based recognizer.
+	 *
+	 * @param speakerModel Speaker recognition model
+	 */
+	fun setSpeakerModel(speakerModel: SpeakerModel)
+
+
+	/**
+	 * Reconfigures recognizer to use grammar
+	 *
+	 * @param grammar      Set of phrases in JSON array of strings or "[]" to use default model graph.
+	 *                     See also vosk_recognizer_new_grm
+	 */
+	fun setGrammar(grammar: String)
+
+	/**
+	 * Configures recognizer to output n-best results
+	 *
+	 * <pre>
+	 *   {
+	 *      "alternatives": [
+	 *          { "text": "one two three four five", "confidence": 0.97 },
+	 *          { "text": "one two three for five", "confidence": 0.03 },
+	 *      ]
+	 *   }
+	 * </pre>
+	 *
+	 * @param maxAlternatives - maximum alternatives to return from recognition results
+	 */
+	fun setMaxAlternatives(maxAlternatives: Int)
+
+	/**
+	 * Enables words with times in the output
+	 *
+	 * <pre>
+	 *   "result" : [{
+	 *       "conf" : 1.000000,
+	 *       "end" : 1.110000,
+	 *       "start" : 0.870000,
+	 *       "word" : "what"
+	 *     }, {
+	 *       "conf" : 1.000000,
+	 *       "end" : 1.530000,
+	 *       "start" : 1.110000,
+	 *       "word" : "zero"
+	 *     }, {
+	 *       "conf" : 1.000000,
+	 *       "end" : 1.950000,
+	 *       "start" : 1.530000,
+	 *       "word" : "zero"
+	 *     }, {
+	 *       "conf" : 1.000000,
+	 *       "end" : 2.340000,
+	 *       "start" : 1.950000,
+	 *       "word" : "zero"
+	 *     }, {
+	 *       "conf" : 1.000000,
+	 *       "end" : 2.610000,
+	 *       "start" : 2.340000,
+	 *       "word" : "one"
+	 *     }],
+	 * </pre>
+	 *
+	 * C equivalent = vosk_recognizer_set_words
+	 * @param words - boolean value
+	 */
+	fun setOutputWordTimes(words: Boolean)
+
+	/**
+	 * Like [setOutputWordTimes] return words and confidences in partial results
+	 *
+	 * @param partialWords - boolean value
+	 */
+	fun setPartialWords(partialWords: Boolean)
+
+	/**
+	 * Set NLSML output
+	 * @param nlsml - boolean value
+	 */
+	fun setNLSML(nlsml: Boolean)
+
+
+	/**
+	 * Set endpointer scaling factor
+	 *
+	 * @param mode Endpointer mode
+	 **/
+	fun setEndPointerMode(mode: EndPointerMode)
+
+	/**
+	 * Set endpointer delays
+	 *
+	 * @param tStartMax timeout for stopping recognition in case of initial silence (usually around 5.0)
+	 * @param tEnd      timeout for stopping recognition in milliseconds after we recognized something (usually around 0.5 - 1.0)
+	 * @param tMax      timeout for forcing utterance end in milliseconds (usually around 20-30)
+	 **/
+	fun setEndPointerDelays(tStartMax: Float, tEnd: Float, tMax: Float)
+
+	/**
+	 * Accept voice data
+	 *
+	 *  accept and process new chunk of voice data
+	 *
+	 *  @param data Audio data in PCM 16-bit mono format.
+	 *  @param length Length of the audio data.
+	 *  @returns
+	 *       1 - If silence is occurred and you can retrieve a new utterance with result method
+	 *       0 - If decoding continues
+	 *      -1 - If exception occurred
+	 */
+	@Throws(AcceptWaveformException::class)
+	fun acceptWaveform(data: ByteArray): Boolean
+
+	/**
+	 * Same as [acceptWaveform] but the version with the short data for language bindings where you have
+	 *  audio as array of shorts
+	 */
+	@Throws(AcceptWaveformException::class)
+	fun acceptWaveform(data: ShortArray): Boolean
+
+	/**
+	 * Same as [acceptWaveform] but the version with the float data for language bindings where you have
+	 *  audio as array of floats
+	 */
+	@Throws(AcceptWaveformException::class)
+	fun acceptWaveform(data: FloatArray): Boolean
+
+	/**
+	 * Returns speech recognition result
+	 *
+	 * @returns the result in JSON format which contains decoded line, decoded
+	 *          words, times in seconds and confidences. You can parse this result
+	 *          with any json parser
+	 *
+	 * <pre>
+	 *  {
+	 *    "text" : "what zero zero zero one"
+	 *  }
+	 * </pre>
+	 *
+	 * If alternatives enabled it returns result with alternatives, see also [setMaxAlternatives].
+	 *
+	 * If word times enabled returns word time, see also [setOutputWordTimes].
+	 */
+	val result: String
+
+	/**
+	 * Returns partial speech recognition
+	 *
+	 * @returns partial speech recognition text which is not yet finalized.
+	 *          result may change as recognizer process more data.
+	 *
+	 * <pre>
+	 * {
+	 *    "partial" : "cyril one eight zero"
+	 * }
+	 * </pre>
+	 */
+	val finalResult: String
+
+	/**
+	 * Returns speech recognition result. Same as result, but doesn't wait for silence
+	 *  You usually call it in the end of the stream to get final bits of audio. It
+	 *  flushes the feature pipeline, so all remaining audio chunks got processed.
+	 *
+	 *  @returns speech result in JSON format.
+	 */
+	val partialResult: String
+
+	/**
+	 * Resets the recognizer
+	 *
+	 *  Resets current results so the recognition can continue from scratch
+	 */
+	fun reset()
+
+	/**
+	 * Releases recognizer object
+	 *
+	 *  Underlying model is also unreferenced and if needed released
+	 */
+	override fun free()
+}

kotlin/src/commonMain/kotlin/org/vosk/SpeakerModel.kt

@@ -0,0 +1,40 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import org.vosk.exception.ModelException
+
+/**
+ * Speaker model is the same as model but contains the data
+ * for speaker identification.
+ *
+ * @since 26 / 12 / 2022
+ * @constructor Loads speaker model data from the file and returns the model object
+ * @param path the path of the model on the filesystem
+ * @throws ModelException if the path provided is invalid
+ */
+expect class SpeakerModel @Throws(ModelException::class) constructor(path: String) : Freeable {
+
+	/**
+	 * Releases the model memory
+	 *
+	 * The model object is reference-counted so if some recognizer
+	 * depends on this model, model might still stay alive. When
+	 * last recognizer is released, model will be released too.
+	 */
+	override fun free()
+}

kotlin/src/commonMain/kotlin/org/vosk/TextProcessor.kt

@@ -0,0 +1,32 @@
+/*
+ * Copyright 2024 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+/**
+ * Inverse text normalization
+ *
+ * @since 2024/06/19
+ * @constructor Create text processor
+ */
+expect class TextProcessor constructor(tagger: Char, verbalizer: Char) : Freeable {
+
+	/** Release text processor */
+	override fun free()
+
+	/** Convert string */
+	fun itn(input: Char): Char
+}

kotlin/src/commonMain/kotlin/org/vosk/Vosk.kt

@@ -0,0 +1,46 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+/**
+ * 26 / 12 / 2022
+ *
+ * Control overarching features of libvosk.
+ */
+expect object Vosk {
+	/**
+	 * Set log level for Kaldi messages
+	 *
+	 *  @param logLevel the level
+	 */
+	fun setLogLevel(logLevel: LogLevel)
+
+	/**
+	 *  Init, automatically select a CUDA device and allow multithreading.
+	 *  Must be called once from the main thread.
+	 *  Has no effect if HAVE_CUDA flag is not set.
+	 */
+	fun gpuInit()
+
+
+	/**
+	 *  Init CUDA device in a multi-threaded environment.
+	 *  Must be called for each thread.
+	 *  Has no effect if HAVE_CUDA flag is not set.
+	 */
+	fun gpuThreadInit()
+}

kotlin/src/commonMain/kotlin/org/vosk/exception/IOException.kt

@@ -0,0 +1,22 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.exception
+
+/**
+ * Internal common IO exception. On JVM this is just a type alias.
+ */
+expect open class IOException(message: String?) : Exception

kotlin/src/commonMain/kotlin/org/vosk/exception/ModelException.kt

@@ -0,0 +1,22 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.exception
+
+/**
+ * Thrown when there is an exception creating a model.
+ */
+class ModelException(path: String): IOException("Failed to find model: $path")

kotlin/src/commonMain/kotlin/org/vosk/exception/RecognizerException.kt

@@ -0,0 +1,22 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.exception
+
+/**
+ * Thrown when the recognizer fails to be created
+ */
+class RecognizerException: IOException("Failed to create recognizer.")

kotlin/src/commonMain/kotlin/org/vosk/json/Alternative.kt

@@ -0,0 +1,31 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.json
+
+import kotlinx.serialization.Serializable
+
+/**
+ * Represents an alternative result.
+ *
+ * @since 26 / 12 / 2022
+ */
+@Serializable
+data class Alternative(
+	val confidence: Double,
+	val result: List<Result> = emptyList(),
+	val text: String
+)

kotlin/src/commonMain/kotlin/org/vosk/json/Extension.kt

@@ -0,0 +1,61 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.json
+
+import kotlinx.serialization.decodeFromString
+import kotlinx.serialization.json.Json
+import kotlinx.serialization.json.encodeToJsonElement
+import org.vosk.Recognizer
+import org.vosk.Model
+import org.vosk.exception.RecognizerException
+
+/*
+ * 26 / 12 / 2022
+ */
+
+/**
+ * Vosk JSON encoder
+ */
+val voskJson = Json { encodeDefaults = true }
+
+/**
+ * Get the result as a JSON object
+ */
+fun Recognizer.resultAsJson(): ResultOutput =
+	voskJson.decodeFromString(result)
+
+/**
+ * Get the final result as a JSON object
+ */
+fun Recognizer.finalResultAsJson(): ResultOutput =
+	voskJson.decodeFromString(finalResult)
+
+/**
+ * Get the partial result as a JSON object
+ */
+fun Recognizer.partialResultAsJson(): PartialResultOutput =
+	voskJson.decodeFromString(partialResult)
+
+
+/**
+ * Create a [Recognizer], but using a list for grammar instead.
+ *
+ * The grammar list is converted into a JSON array
+ */
+@Throws(RecognizerException::class)
+fun Recognizer(model: Model, sampleRate: Float, grammar: List<String>) =
+	Recognizer(model, sampleRate, voskJson.encodeToJsonElement(grammar).toString())

kotlin/src/commonMain/kotlin/org/vosk/json/PartialResultOutput.kt

@@ -0,0 +1,32 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.json
+
+import kotlinx.serialization.SerialName
+import kotlinx.serialization.Serializable
+
+/**
+ * Represents a partial result
+ *
+ * @since 26 / 12 / 2022
+ */
+@Serializable
+data class PartialResultOutput(
+	val partial: String,
+	@SerialName("partial_result")
+	val partialResult: List<Result> = emptyList(),
+)

kotlin/src/commonMain/kotlin/org/vosk/json/Result.kt

@@ -0,0 +1,32 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.json
+
+import kotlinx.serialization.Serializable
+
+/**
+ * Represents a result for any given word.
+ *
+ * @since 26 / 12 / 2022
+ */
+@Serializable
+data class Result(
+	val conf: Double? = null,
+	val end: Double,
+	val start: Double,
+	val word: String,
+)

kotlin/src/commonMain/kotlin/org/vosk/json/ResultOutput.kt

@@ -0,0 +1,31 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.json
+
+import kotlinx.serialization.Serializable
+
+/**
+ * Result output
+ *
+ * @since 26 / 12 / 2022
+ */
+@Serializable
+data class ResultOutput(
+	val alternatives: List<Alternative> = emptyList(),
+	val result: List<Result> = emptyList(),
+	val text: String? = null
+)

kotlin/src/commonMain/kotlin/org/vosk/json/WaveformResult.kt

@@ -0,0 +1,38 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.json
+
+/**
+ * For extension functions transforming (input streams->recognizers) into flows.
+ */
+sealed interface WaveformResult {
+
+	/**
+	 * A result made after a period of silence.
+	 */
+	data class Result(val result: ResultOutput) : WaveformResult
+
+	/**
+	 * A partial result that is being made in progress.
+	 */
+	data class PartialResult(val result: PartialResultOutput) : WaveformResult
+
+	/**
+	 * A final result made after there is no more content to feed.
+	 */
+	data class FinalResult(val result: ResultOutput) : WaveformResult
+}

kotlin/src/jvmMain/kotlin/org/vosk/BatchModel.kt

@@ -0,0 +1,83 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import com.sun.jna.PointerType
+import org.vosk.exception.ModelException
+import java.io.File
+import java.nio.file.Path
+import kotlin.io.path.absolutePathString
+
+
+/**
+ * Batch model object
+ *
+ * @since 26 / 12 / 2022
+ */
+actual class BatchModel : Freeable, PointerType, AutoCloseable {
+
+	/**
+	 * Empty constructor for JNA
+	 */
+	constructor()
+
+	/**
+	 * Creates the batch recognizer object
+	 */
+	@Throws(ModelException::class)
+	actual constructor(path: String) : super(
+		LibVosk.vosk_batch_model_new(path) ?: throw ModelException(path)
+	)
+
+	/**
+	 * Constructor using a Path, will retrieve absolutePath
+	 *
+	 * @param path to batch model
+	 */
+	@Throws(ModelException::class)
+	constructor(path: Path) : this(path.absolutePathString())
+
+	/**
+	 * Constructor using a File, will retrieve absolutePath
+	 *
+	 * @param file to batch model
+	 */
+	@Throws(ModelException::class)
+	constructor(file: File) : this(file.absolutePath)
+
+	/**
+	 * Releases batch model object
+	 */
+	actual override fun free() {
+		LibVosk.vosk_batch_model_free(this)
+	}
+
+	/**
+	 * Wait for the processing
+	 */
+	actual fun await() {
+		LibVosk.vosk_batch_model_wait(this)
+	}
+
+	/**
+	 * @see free
+	 */
+	override fun close() {
+		free()
+	}
+
+}

kotlin/src/jvmMain/kotlin/org/vosk/BatchRecognizer.kt

@@ -0,0 +1,95 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import com.sun.jna.PointerType
+
+
+/**
+ * Batch recognizer object
+ *
+ * @since 26 / 12 / 2022
+ * @constructor Creates batch recognizer object
+ */
+actual class BatchRecognizer : Freeable, PointerType, AutoCloseable {
+
+	/**
+	 * Empty constructor for JNA
+	 */
+	constructor()
+
+	/**
+	 * Creates batch recognizer object
+	 */
+	actual constructor(model: BatchModel, sampleRate: Float) :
+			super(LibVosk.vosk_batch_recognizer_new(model, sampleRate))
+
+	/**
+	 * Releases batch recognizer object
+	 */
+	actual override fun free() {
+		LibVosk.vosk_batch_recognizer_free(this);
+	}
+
+	/**
+	 * Accept batch voice data
+	 */
+	actual fun acceptWaveform(data: ByteArray) {
+		LibVosk.vosk_batch_recognizer_accept_waveform(this, data, data.size)
+	}
+
+	/**
+	 * Set NLSML output
+	 * @param nlsml - boolean value
+	 */
+	actual fun setNLSML(nlsml: Boolean) {
+		LibVosk.vosk_batch_recognizer_set_nlsml(this, nlsml)
+	}
+
+	/**
+	 *  Closes the stream
+	 */
+	actual fun finishStream() {
+		LibVosk.vosk_batch_recognizer_finish_stream(this)
+	}
+
+	/**
+	 * Return results
+	 */
+	actual val frontResult: String
+		get() = LibVosk.vosk_batch_recognizer_front_result(this)
+
+	/**
+	 * Release and free first retrieved result
+	 */
+	actual fun pop() {
+		LibVosk.vosk_batch_recognizer_pop(this)
+	}
+
+	/**
+	 * Get amount of pending chunks for more intelligent waiting
+	 */
+	actual val pendingChunks: Int
+		get() = LibVosk.vosk_batch_recognizer_get_pending_chunks(this)
+
+	/**
+	 * @see free
+	 */
+	override fun close() {
+		free()
+	}
+}

kotlin/src/jvmMain/kotlin/org/vosk/LibVosk.kt

@@ -0,0 +1,213 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import com.sun.jna.Native
+import com.sun.jna.Platform
+import com.sun.jna.Pointer
+import java.io.File
+import java.io.IOException
+import java.io.InputStream
+import java.nio.file.Files
+import java.nio.file.StandardCopyOption
+
+/**
+ * 26 / 12 / 2022
+ */
+@Suppress("FunctionName")
+internal object LibVosk {
+
+	@JvmStatic
+	@Deprecated(
+		"LibVosk is now for internal JNA, use Vosk instead",
+		ReplaceWith("Vosk.setLogLevel(logLevel)", "org.vosk.Vosk")
+	)
+	fun setLogLevel(logLevel: LogLevel) {
+		Vosk.setLogLevel(logLevel)
+	}
+
+	@Throws(IOException::class)
+	private fun unpackDll(targetDir: File, lib: String) {
+		Vosk::class.java.getResourceAsStream("/win32-x86-64/$lib.dll")!!.use {
+			Files.copy(
+				it,
+				File(targetDir, "$lib.dll").toPath(),
+				StandardCopyOption.REPLACE_EXISTING
+			)
+		}
+	}
+
+	init {
+		when {
+			Platform.isAndroid() -> {
+				Native.register(LibVosk::class.java, "vosk")
+			}
+
+			Platform.isWindows() -> {
+				// We have to unpack dependencies
+				try {
+					// To get a tmp folder we unpack small library and mark it for deletion
+					val tmpFile: File = Native.extractFromResourcePath(
+						"/win32-x86-64/empty",
+						LibVosk::class.java.classLoader
+					)
+					val tmpDir = tmpFile.parentFile!!
+					File(tmpDir, tmpFile.name + ".x").createNewFile()
+
+					// Now unpack dependencies
+					unpackDll(tmpDir, "libwinpthread-1");
+					unpackDll(tmpDir, "libgcc_s_seh-1");
+					unpackDll(tmpDir, "libstdc++-6");
+
+				} catch (e: IOException) {
+					// Nothing for now, it will fail on next step
+				} finally {
+					Native.register(LibVosk::class.java, "libvosk");
+				}
+			}
+
+			else -> {
+				Native.register(LibVosk::class.java, "vosk");
+			}
+		}
+	}
+
+
+	external fun vosk_model_new(path: String): Pointer?
+
+	external fun vosk_model_free(model: Model)
+
+	external fun vosk_model_find_word(model: Model, word: String): Int
+
+
+	external fun vosk_spk_model_new(path: String): Pointer?
+
+	external fun vosk_spk_model_free(model: SpeakerModel)
+
+
+	external fun vosk_recognizer_new(model: Model, sampleRate: Float): Pointer?
+
+	external fun vosk_recognizer_new_spk(
+		model: Model,
+		sampleRate: Float,
+		spkModel: SpeakerModel
+	): Pointer?
+
+	external fun vosk_recognizer_new_grm(
+		model: Model,
+		sampleRate: Float,
+		grammar: String?
+	): Pointer?
+
+	external fun vosk_recognizer_set_spk_model(recognizer: Recognizer, spk_model: SpeakerModel)
+
+	external fun vosk_recognizer_set_grm(recognizer: Recognizer, grammar: String)
+
+	external fun vosk_recognizer_set_max_alternatives(recognizer: Recognizer, maxAlternatives: Int)
+
+	external fun vosk_recognizer_set_words(recognizer: Recognizer, words: Boolean)
+
+	external fun vosk_recognizer_set_partial_words(recognizer: Recognizer, partial_words: Boolean)
+
+	external fun vosk_recognizer_set_nlsml(recognizer: Recognizer, nlsml: Boolean)
+
+
+	external fun vosk_recognizer_accept_waveform(
+		recognizer: Recognizer,
+		data: ByteArray?,
+		len: Int
+	): Int
+
+	external fun vosk_recognizer_accept_waveform_s(
+		recognizer: Recognizer,
+		data: ShortArray?,
+		len: Int
+	): Int
+
+	external fun vosk_recognizer_accept_waveform_f(
+		recognizer: Recognizer,
+		data: FloatArray?,
+		len: Int
+	): Int
+
+
+	external fun vosk_recognizer_result(recognizer: Recognizer): String
+
+	external fun vosk_recognizer_final_result(recognizer: Recognizer): String
+
+	external fun vosk_recognizer_partial_result(recognizer: Recognizer): String
+
+
+	external fun vosk_recognizer_reset(recognizer: Recognizer)
+
+	external fun vosk_recognizer_free(recognizer: Recognizer)
+
+	external fun vosk_set_log_level(level: Int)
+
+	external fun vosk_gpu_init()
+
+	external fun vosk_gpu_thread_init()
+
+
+	external fun vosk_batch_model_new(path: String): Pointer?
+
+	external fun vosk_batch_model_free(model: BatchModel)
+
+	external fun vosk_batch_model_wait(model: BatchModel)
+
+	external fun vosk_batch_recognizer_new(batchModel: BatchModel, sampleRate: Float): Pointer
+
+	external fun vosk_batch_recognizer_free(recognizer: BatchRecognizer)
+
+	external fun vosk_batch_recognizer_accept_waveform(
+		recognizer: BatchRecognizer,
+		data: ByteArray?,
+		length: Int
+	)
+
+	external fun vosk_batch_recognizer_set_nlsml(
+		recognizer: BatchRecognizer,
+		nlsml: Boolean
+	)
+
+	external fun vosk_batch_recognizer_finish_stream(
+		recognizer: BatchRecognizer
+	)
+
+	external fun vosk_batch_recognizer_front_result(
+		recognizer: BatchRecognizer
+	): String
+
+	external fun vosk_batch_recognizer_pop(recognizer: BatchRecognizer)
+
+	external fun vosk_batch_recognizer_get_pending_chunks(recognizer: BatchRecognizer): Int
+
+	external fun vosk_text_processor_new(tagger: Char, verbalizer: Char): Pointer
+
+	external fun vosk_text_processor_free(processor: TextProcessor)
+
+	external fun vosk_text_processor_itn(processor: TextProcessor, input: Char): Char
+
+	external fun vosk_recognizer_set_endpointer_mode(recognizer: Recognizer, ordinal: Int)
+
+	external fun vosk_recognizer_set_endpointer_delays(
+		recognizer: Recognizer,
+		tStartMax: Float,
+		tEnd: Float,
+		tMax: Float
+	)
+}

kotlin/src/jvmMain/kotlin/org/vosk/Model.kt

@@ -0,0 +1,95 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import com.sun.jna.PointerType
+import org.vosk.exception.IOException
+import org.vosk.exception.ModelException
+import java.io.File
+import java.nio.file.Path
+import kotlin.io.path.absolutePathString
+
+
+/**
+ * Model stores all the data required for recognition
+ *
+ * It contains static data and can be shared across processing
+ * threads.
+ *
+ * @since 26 / 12 / 2022
+ * @constructor Loads model data from the file and returns the model object
+ * @throws IOException if the path provided is invalid
+ */
+actual class Model : Freeable, PointerType, AutoCloseable {
+
+	/**
+	 * Empty constructor for JNA
+	 */
+	constructor()
+
+	/**
+	 * Loads model data from the file and returns the model object
+	 */
+	@Throws(IOException::class)
+	actual constructor(path: String) : super(
+		LibVosk.vosk_model_new(path) ?: throw ModelException(path)
+	)
+
+	/**
+	 * Constructor using a Path, will retrieve absolutePath
+	 *
+	 * @param path to batch model
+	 */
+	@Throws(IOException::class)
+	constructor(path: Path) : this(path.absolutePathString())
+
+	/**
+	 * Constructor using a File, will retrieve absolutePath
+	 *
+	 * @param file to batch model
+	 */
+	@Throws(IOException::class)
+	constructor(file: File) : this(file.absolutePath)
+
+	/**
+	 * Check if a word can be recognized by the model
+	 * @param word: the word
+	 * @returns the word symbol if @param word exists inside the model
+	 * or -1 otherwise.
+	 * Reminding that word symbol 0 is for <epsilon>
+	 */
+	actual fun findWord(word: String): Int =
+		LibVosk.vosk_model_find_word(this, word)
+
+	/**
+	 * Releases the model memory
+	 *
+	 *  The model object is reference-counted so if some recognizer
+	 *  depends on this model, model might still stay alive. When
+	 *  last recognizer is released, model will be released too.
+	 */
+	actual override fun free() {
+		LibVosk.vosk_model_free(this)
+	}
+
+	/**
+	 * @see free
+	 */
+	override fun close() {
+		free()
+	}
+}

kotlin/src/jvmMain/kotlin/org/vosk/Recognizer.kt

@@ -0,0 +1,353 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import com.sun.jna.PointerType
+import org.vosk.exception.RecognizerException
+
+/**
+ * Recognizer object is the main object which processes data.
+ *
+ * Each recognizer usually runs in own thread and takes audio as input.
+ * Once audio is processed recognizer returns JSON object as a string
+ * which represent decoded information - words, confidences, times, n-best lists,
+ * speaker information and so on
+ *
+ * @since 26 / 12 / 2022
+ */
+actual class Recognizer : Freeable, PointerType, AutoCloseable {
+
+	/**
+	 * Empty constructor for JNA
+	 */
+	constructor()
+
+	/**
+	 * Creates the recognizer object
+	 *
+	 *  The recognizers process the speech and return text using shared model data
+	 *  @param model       VoskModel containing static data for recognizer. Model can be
+	 *                     shared across recognizers, even running in different threads.
+	 *  @param sampleRate The sample rate of the audio you going to feed into the recognizer.
+	 *                     Make sure this rate matches the audio content, it is a common
+	 *                     issue causing accuracy problems.
+	 * @throws RecognizerException if a problem occurred
+	 */
+	@Throws(RecognizerException::class)
+	actual constructor(model: Model, sampleRate: Float) :
+			super(LibVosk.vosk_recognizer_new(model, sampleRate) ?: throw RecognizerException())
+
+	/**
+	 * Creates the recognizer object with speaker recognition
+	 *
+	 *  With the speaker recognition mode the recognizer not just recognize
+	 *  text but also return speaker vectors one can use for speaker identification
+	 *
+	 *  @param model       VoskModel containing static data for recognizer. Model can be
+	 *                     shared across recognizers, even running in different threads.
+	 *  @param sampleRate The sample rate of the audio you going to feed into the recognizer.
+	 *                     Make sure this rate matches the audio content, it is a common
+	 *                     issue causing accuracy problems.
+	 *  @param speakerModel speaker model for speaker identification
+	 * @throws RecognizerException if a problem occurred
+	 */
+	@Throws(RecognizerException::class)
+	actual constructor(
+		model: Model,
+		sampleRate: Float,
+		speakerModel: SpeakerModel
+	) : super(
+		LibVosk.vosk_recognizer_new_spk(model, sampleRate, speakerModel)
+			?: throw RecognizerException()
+	)
+
+	/**
+	 * Creates the recognizer object with the phrase list
+	 *
+	 *  Sometimes when you want to improve recognition accuracy and when you don't need
+	 *  to recognize large vocabulary you can specify a list of phrases to recognize. This
+	 *  will improve recognizer speed and accuracy but might return [unk] if user said
+	 *  something different.
+	 *
+	 *  Only recognizers with lookahead models support this type of quick configuration.
+	 *  Precompiled HCLG graph models are not supported.
+	 *
+	 *  @param model       VoskModel containing static data for recognizer. Model can be
+	 *                     shared across recognizers, even running in different threads.
+	 *  @param sampleRate The valuesample rate of the audio you going to feed into the recognizer.
+	 *                     Make sure this rate matches the audio content, it is a common
+	 *                     issue causing accuracy problems.
+	 *  @param grammar The string with the list of phrases to recognize as JSON array of strings,
+	 *                 for example "["one two three four five", "[unk]"]".
+	 *
+	 * @throws RecognizerException if a problem occurred
+	 */
+	@Throws(RecognizerException::class)
+	actual constructor(
+		model: Model,
+		sampleRate: Float,
+		grammar: String
+	) : super(
+		LibVosk.vosk_recognizer_new_grm(model, sampleRate, grammar) ?: throw RecognizerException()
+	)
+
+	/**
+	 * JVM analog of Kotlin extension
+	 * @see [org.vosk.json.Recognizer]
+	 */
+	@Throws(RecognizerException::class)
+	constructor(
+		model: Model,
+		sampleRate: Float,
+		grammar: List<String>
+	) : super(
+		// We need the full qualifer to avoid any build issues
+		@Suppress("RemoveRedundantQualifierName")
+		org.vosk.json.Recognizer(model, sampleRate, grammar).pointer
+	)
+
+	/**
+	 * Adds speaker model to already initialized recognizer
+	 *
+	 * Can add speaker recognition model to already created recognizer. Helps to initialize
+	 * speaker recognition for grammar-based recognizer.
+	 *
+	 * @param speakerModel Speaker recognition model
+	 */
+	actual fun setSpeakerModel(speakerModel: SpeakerModel) {
+		LibVosk.vosk_recognizer_set_spk_model(this, speakerModel)
+	}
+
+	/**
+	 * Reconfigures recognizer to use grammar
+	 *
+	 * @param grammar      Set of phrases in JSON array of strings or "[]" to use default model graph.
+	 *                     See also vosk_recognizer_new_grm
+	 */
+	actual fun setGrammar(grammar: String) {
+		LibVosk.vosk_recognizer_set_grm(this, grammar)
+	}
+
+	/**
+	 * Configures recognizer to output n-best results
+	 *
+	 * <pre>
+	 *   {
+	 *      "alternatives": [
+	 *          { "text": "one two three four five", "confidence": 0.97 },
+	 *          { "text": "one two three for five", "confidence": 0.03 },
+	 *      ]
+	 *   }
+	 * </pre>
+	 *
+	 * @param maxAlternatives - maximum alternatives to return from recognition results
+	 */
+	actual fun setMaxAlternatives(maxAlternatives: Int) {
+		LibVosk.vosk_recognizer_set_max_alternatives(this, maxAlternatives)
+	}
+
+	/**
+	 * Enables words with times in the output
+	 *
+	 * <pre>
+	 *   "result" : [{
+	 *       "conf" : 1.000000,
+	 *       "end" : 1.110000,
+	 *       "start" : 0.870000,
+	 *       "word" : "what"
+	 *     }, {
+	 *       "conf" : 1.000000,
+	 *       "end" : 1.530000,
+	 *       "start" : 1.110000,
+	 *       "word" : "zero"
+	 *     }, {
+	 *       "conf" : 1.000000,
+	 *       "end" : 1.950000,
+	 *       "start" : 1.530000,
+	 *       "word" : "zero"
+	 *     }, {
+	 *       "conf" : 1.000000,
+	 *       "end" : 2.340000,
+	 *       "start" : 1.950000,
+	 *       "word" : "zero"
+	 *     }, {
+	 *       "conf" : 1.000000,
+	 *       "end" : 2.610000,
+	 *       "start" : 2.340000,
+	 *       "word" : "one"
+	 *     }],
+	 * </pre>
+	 *
+	 * @param words - boolean value
+	 */
+	actual fun setOutputWordTimes(words: Boolean) {
+		LibVosk.vosk_recognizer_set_words(this, words)
+	}
+
+	/**
+	 * Like [setOutputWordTimes] return words and confidences in partial results
+	 *
+	 * @param partialWords - boolean value
+	 */
+	actual fun setPartialWords(partialWords: Boolean) {
+		LibVosk.vosk_recognizer_set_partial_words(this, partialWords)
+	}
+
+	/**
+	 * Set NLSML output
+	 * @param nlsml - boolean value
+	 */
+	actual fun setNLSML(nlsml: Boolean) {
+		LibVosk.vosk_recognizer_set_nlsml(this, nlsml)
+	}
+
+	/**
+	 * Accept voice data
+	 *
+	 *  accept and process new chunk of voice data
+	 *
+	 *  @param data Audio data in PCM 16-bit mono format.
+	 *  @param length Length of the audio data.
+	 *  @returns
+	 *       1 - If silence is occurred and you can retrieve a new utterance with result method
+	 *       0 - If decoding continues
+	 *      -1 - If exception occurred
+	 */
+	@Throws(AcceptWaveformException::class)
+	actual fun acceptWaveform(data: ByteArray): Boolean {
+		val result = LibVosk.vosk_recognizer_accept_waveform(this, data, data.size)
+		if (result == -1) throw AcceptWaveformException(data)
+		return result == 1
+	}
+
+	/**
+	 * Same as [acceptWaveform] but the version with the short data for language bindings where you have
+	 *  audio as array of shorts
+	 */
+	@Throws(AcceptWaveformException::class)
+	actual fun acceptWaveform(data: ShortArray): Boolean {
+		val result = LibVosk.vosk_recognizer_accept_waveform_s(this, data, data.size)
+		if (result == -1) throw AcceptWaveformException(data)
+		return result == 1
+	}
+
+	/**
+	 * Same as [acceptWaveform] but the version with the float data for language bindings where you have
+	 *  audio as array of floats
+	 */
+	@Throws(AcceptWaveformException::class)
+	actual fun acceptWaveform(data: FloatArray): Boolean {
+		val result = LibVosk.vosk_recognizer_accept_waveform_f(this, data, data.size)
+		if (result == -1) throw AcceptWaveformException(data)
+		return result == 1
+	}
+
+
+	/**
+	 * Returns speech recognition result
+	 *
+	 * @returns the result in JSON format which contains decoded line, decoded
+	 *          words, times in seconds and confidences. You can parse this result
+	 *          with any json parser
+	 *
+	 * <pre>
+	 *  {
+	 *    "text" : "what zero zero zero one"
+	 *  }
+	 * </pre>
+	 *
+	 * If alternatives enabled it returns result with alternatives, see also [setMaxAlternatives].
+	 *
+	 * If word times enabled returns word time, see also [setOutputWordTimes].
+	 */
+	actual val result: String
+		get() = LibVosk.vosk_recognizer_result(this)
+
+	/**
+	 * Returns partial speech recognition
+	 *
+	 * @returns partial speech recognition text which is not yet finalized.
+	 *          result may change as recognizer process more data.
+	 *
+	 * <pre>
+	 * {
+	 *    "partial" : "cyril one eight zero"
+	 * }
+	 * </pre>
+	 */
+	actual val finalResult: String
+		get() = LibVosk.vosk_recognizer_final_result(this)
+
+	/**
+	 * Returns speech recognition result. Same as result, but doesn't wait for silence
+	 *  You usually call it in the end of the stream to get final bits of audio. It
+	 *  flushes the feature pipeline, so all remaining audio chunks got processed.
+	 *
+	 *  @returns speech result in JSON format.
+	 */
+	actual val partialResult: String
+		get() = LibVosk.vosk_recognizer_partial_result(this)
+
+	/**
+	 * Resets the recognizer
+	 *
+	 *  Resets current results so the recognition can continue from scratch
+	 */
+	actual fun reset() {
+		LibVosk.vosk_recognizer_reset(this)
+	}
+
+	/**
+	 * Releases recognizer object
+	 *
+	 *  Underlying model is also unreferenced and if needed released
+	 */
+	actual override fun free() {
+		LibVosk.vosk_recognizer_free(this)
+	}
+
+	/**
+	 * @see free
+	 */
+	override fun close() {
+		free()
+	}
+
+	/**
+	 * Set endpointer scaling factor
+	 *
+	 * @param mode Endpointer mode
+	 **/
+	actual fun setEndPointerMode(mode: EndPointerMode) {
+		LibVosk.vosk_recognizer_set_endpointer_mode(this, mode.ordinal)
+	}
+
+	/**
+	 * Set endpointer delays
+	 *
+	 * @param tStartMax timeout for stopping recognition in case of initial silence (usually around 5.0)
+	 * @param tEnd      timeout for stopping recognition in milliseconds after we recognized something (usually around 0.5 - 1.0)
+	 * @param tMax      timeout for forcing utterance end in milliseconds (usually around 20-30)
+	 **/
+	actual fun setEndPointerDelays(
+		tStartMax: Float,
+		tEnd: Float,
+		tMax: Float
+	) {
+		LibVosk.vosk_recognizer_set_endpointer_delays(this, tStartMax, tEnd, tMax)
+	}
+}

kotlin/src/jvmMain/kotlin/org/vosk/SpeakerModel.kt

@@ -0,0 +1,84 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import com.sun.jna.PointerType
+import org.vosk.exception.ModelException
+import java.io.File
+import java.nio.file.Path
+import kotlin.io.path.absolutePathString
+
+/**
+ * Speaker model is the same as model but contains the data
+ * for speaker identification.
+ *
+ * @since 26 / 12 / 2022
+ * @constructor Loads speaker model data from the file and returns the model object
+ * @throws ModelException if the path provided is invalid
+ */
+actual class SpeakerModel : Freeable, PointerType, AutoCloseable {
+
+	/**
+	 * Empty constructor for JNA
+	 */
+	constructor()
+
+	/**
+	 * Loads speaker model data from the file and returns the model object
+	 *
+	 * @param path the path of the model on the filesystem
+	 */
+	@Throws(ModelException::class)
+	actual constructor(path: String) : super(
+		LibVosk.vosk_spk_model_new(path) ?: throw ModelException(path)
+	)
+
+	/**
+	 * Constructor using a Path, will retrieve absolutePath
+	 *
+	 * @param path to batch model
+	 */
+	@Throws(ModelException::class)
+	constructor(path: Path) : this(path.absolutePathString())
+
+	/**
+	 * Constructor using a File, will retrieve absolutePath
+	 *
+	 * @param file to batch model
+	 */
+	@Throws(ModelException::class)
+	constructor(file: File) : this(file.absolutePath)
+
+	/**
+	 * Releases the model memory
+	 *
+	 * The model object is reference-counted so if some recognizer
+	 * depends on this model, model might still stay alive. When
+	 * last recognizer is released, model will be released too.
+	 */
+	actual override fun free() {
+		LibVosk.vosk_spk_model_free(this)
+	}
+
+	/**
+	 * @see free
+	 */
+	override fun close() {
+		free()
+	}
+
+}

kotlin/src/jvmMain/kotlin/org/vosk/TextProcessor.kt

@@ -0,0 +1,51 @@
+/*
+ * Copyright 2024 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import com.sun.jna.PointerType
+
+/**
+ * Inverse text normalization
+ *
+ * @since 2024/06/19
+ */
+actual class TextProcessor :
+	Freeable, PointerType, AutoCloseable {
+
+	/**
+	 * Create text processor
+	 */
+	actual constructor(tagger: Char, verbalizer: Char) :
+			super(LibVosk.vosk_text_processor_new(tagger, verbalizer))
+
+
+	/** Release text processor */
+	actual override fun free() {
+		LibVosk.vosk_text_processor_free(this)
+	}
+
+	/** Convert string */
+	actual fun itn(input: Char): Char =
+		LibVosk.vosk_text_processor_itn(this, input)
+
+	/**
+	 * @see free
+	 */
+	override fun close() {
+		free()
+	}
+}

kotlin/src/jvmMain/kotlin/org/vosk/Vosk.kt

@@ -0,0 +1,55 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+
+/**
+ * 26 / 12 / 2022
+ *
+ * Control overarching features of libvosk.
+ */
+actual object Vosk {
+	/**
+	 * Set log level for Kaldi messages
+	 *
+	 *  @param logLevel the level
+	 */
+	@JvmStatic
+	actual fun setLogLevel(logLevel: LogLevel) {
+		LibVosk.vosk_set_log_level(logLevel.value)
+	}
+
+	/**
+	 *  Init, automatically select a CUDA device and allow multithreading.
+	 *  Must be called once from the main thread.
+	 *  Has no effect if HAVE_CUDA flag is not set.
+	 */
+	@JvmStatic
+	actual fun gpuInit() {
+		LibVosk.vosk_gpu_init()
+	}
+
+	/**
+	 *  Init CUDA device in a multi-threaded environment.
+	 *  Must be called for each thread.
+	 *  Has no effect if HAVE_CUDA flag is not set.
+	 */
+	@JvmStatic
+	actual fun gpuThreadInit() {
+		LibVosk.vosk_gpu_thread_init()
+	}
+}

kotlin/src/jvmMain/kotlin/org/vosk/WaveformExt.kt

@@ -0,0 +1,55 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import kotlinx.coroutines.Dispatchers
+import kotlinx.coroutines.flow.Flow
+import kotlinx.coroutines.flow.flowOn
+import kotlinx.coroutines.flow.map
+import org.vosk.json.WaveformResult
+import org.vosk.json.finalResultAsJson
+import org.vosk.json.partialResultAsJson
+import org.vosk.json.resultAsJson
+
+/**
+ * Feed an [Flow] of [ByteArray] into a [Recognizer].
+ *
+ * The returned flow will emit an [WaveformResult] for each result parsed.
+ *
+ * This expects a null terminator to signify the end of a stream.
+ *
+ * This will not close the [recognizer] at the end of reading.
+ *
+ * Any exceptions will be fed into the flow,
+ *  and should be collected via [kotlinx.coroutines.flow.catch].
+ *
+ * Flows on [Dispatchers.IO] to prevent blocking the main thread.
+ *
+ * @see [Recognizer.acceptWaveform]
+ */
+fun Flow<ByteArray?>.feed(recognizer: Recognizer): Flow<WaveformResult> =
+	map {
+		if (it != null) {
+			if (recognizer.acceptWaveform(it)) {
+				WaveformResult.Result(recognizer.resultAsJson())
+			} else {
+				WaveformResult.PartialResult(recognizer.partialResultAsJson())
+			}
+		} else {
+			WaveformResult.FinalResult(recognizer.finalResultAsJson())
+		}
+	}.flowOn(Dispatchers.IO)

kotlin/src/jvmMain/kotlin/org/vosk/exception/IOException.kt

@@ -0,0 +1,22 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk.exception
+
+/**
+ * Analog of [java.io.IOException]
+ */
+actual typealias IOException = java.io.IOException

kotlin/src/jvmTest/kotlin/DecoderTest.kt

@@ -0,0 +1,212 @@
+/*
+ * Copyright 2023 Alpha Cephei Inc.
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+import kotlinx.coroutines.*
+import kotlinx.coroutines.flow.*
+import kotlinx.coroutines.test.runTest
+import org.vosk.*
+import java.io.BufferedInputStream
+import java.io.FileInputStream
+import java.io.IOException
+import java.nio.ByteBuffer
+import java.nio.ByteOrder
+import javax.sound.sampled.AudioSystem
+import javax.sound.sampled.UnsupportedAudioFileException
+import kotlin.test.Test
+
+class DecoderTest {
+	val modelPath = System.getenv("MODEL")
+	val testFile = System.getenv("AUDIO")
+
+	init {
+		System.load(System.getenv("LIBRARY"))
+		Vosk.setLogLevel(LogLevel.DEBUG)
+	}
+
+	@Test
+	fun grammarList() {
+		Model(modelPath).use { model ->
+			Recognizer(model, 16000f, listOf("one")).apply {
+				setMaxAlternatives(10)
+				setOutputWordTimes(true)
+				setPartialWords(true)
+			}
+		}
+	}
+
+	@Test
+	@Throws(IOException::class, UnsupportedAudioFileException::class)
+	fun decoderTest() {
+		Model(modelPath).use { model ->
+			AudioSystem.getAudioInputStream(BufferedInputStream(FileInputStream(testFile)))
+				.use { ais ->
+					Recognizer(model, 16000f).apply {
+						setMaxAlternatives(10)
+						setOutputWordTimes(true)
+						setPartialWords(true)
+					}.use { recognizer ->
+						val b = ByteArray(4096)
+						while (ais.read(b) >= 0) {
+							if (recognizer.acceptWaveform(b)) {
+								println(recognizer.result)
+							} else {
+								println(recognizer.partialResult)
+							}
+						}
+						println(recognizer.finalResult)
+					}
+				}
+		}
+	}
+
+	@OptIn(ExperimentalCoroutinesApi::class)
+	@Test
+	@Throws(IOException::class, UnsupportedAudioFileException::class)
+	fun decoderTestFlow() = runTest {
+		Model(modelPath).use { model ->
+			Recognizer(model, 16000f).apply {
+				setMaxAlternatives(10)
+				setOutputWordTimes(true)
+				setPartialWords(true)
+			}.use { recognizer ->
+				AudioSystem.getAudioInputStream(BufferedInputStream(FileInputStream(testFile)))
+					.use { ais ->
+						flow {
+							val b = ByteArray(4096)
+							while (ais.read(b) >= 0) {
+								emit(b)
+							}
+							emit(null)
+						}.flowOn(Dispatchers.IO)
+							.feed(recognizer)
+							.collect {
+								println(it)
+							}
+					}
+			}
+		}
+	}
+
+	/**
+	 * This test aims to simulate the situation for Dicio,
+	 *  In which we can receive the audio input stream before the recognizer is setup.
+	 *
+	 *  It is recommended to use a large model on desktop to properly see how long it takes to load up.
+	 */
+	@Test
+	@Throws(IOException::class, UnsupportedAudioFileException::class)
+	fun decoderTestFlowBuffered() {
+		// Scope to buffer into
+		val scope = CoroutineScope(Dispatchers.IO)
+
+		AudioSystem.getAudioInputStream(BufferedInputStream(FileInputStream(testFile))).use { ais ->
+			val byteFlow = flow {
+				val b = ByteArray(4096)
+				while (ais.read(b) >= 0) {
+					emit(b)
+				}
+				emit(null)
+			}.flowOn(Dispatchers.IO)
+				.shareIn(scope, SharingStarted.Eagerly, 100)
+
+			// Tell us the current buffer size
+			println("Buffered size:" + byteFlow.replayCache.size)
+
+			var startTime = System.currentTimeMillis()
+
+			Model(modelPath).use { model ->
+				var resultTime = System.currentTimeMillis() - startTime
+				println("Model initialized in: $resultTime")
+				startTime = System.currentTimeMillis()
+
+				Recognizer(model, 16000f).apply {
+					setMaxAlternatives(10)
+					setOutputWordTimes(true)
+					setPartialWords(true)
+				}.use { recognizer ->
+					resultTime = System.currentTimeMillis() - startTime
+					println("Recognizer initialized in: $resultTime")
+					println("Buffered size:" + byteFlow.replayCache.size)
+
+					runBlocking {
+						byteFlow
+							.feed(recognizer)
+							.take(byteFlow.replayCache.size)
+							.collect {
+								println(it)
+							}
+					}
+				}
+			}
+		}
+	}
+
+	@Test
+	@Throws(IOException::class, UnsupportedAudioFileException::class)
+	fun decoderTestShort() {
+		Model(modelPath).use { model ->
+			AudioSystem.getAudioInputStream(BufferedInputStream(FileInputStream(testFile)))
+				.use { ais ->
+					Recognizer(model, 16000f).use { recognizer ->
+						val b = ByteArray(4096)
+						val s = ShortArray(2048)
+						while (ais.read(b) >= 0) {
+							ByteBuffer.wrap(b).order(ByteOrder.LITTLE_ENDIAN).asShortBuffer().get(s)
+							if (recognizer.acceptWaveform(s)) {
+								println(recognizer.result)
+							} else {
+								println(recognizer.partialResult)
+							}
+						}
+						println(recognizer.finalResult)
+					}
+				}
+		}
+	}
+
+	@Test
+	@Throws(IOException::class, UnsupportedAudioFileException::class)
+	fun decoderTestGrammar() {
+		Model(modelPath).use { model ->
+			AudioSystem.getAudioInputStream(BufferedInputStream(FileInputStream(testFile)))
+				.use { ais ->
+					Recognizer(
+						model, 16000f, "[\"one two three four five six seven eight nine zero oh\"]"
+					).use { recognizer ->
+						val b = ByteArray(4096)
+						while (ais.read(b) >= 0) {
+							if (recognizer.acceptWaveform(b)) {
+								println(recognizer.result)
+							} else {
+								println(recognizer.partialResult)
+							}
+						}
+						println(recognizer.finalResult)
+					}
+				}
+		}
+	}
+
+	@Test
+	fun decoderTestException() {
+		try {
+			val model = Model("model_missing")
+			assert(false)
+		} catch (e: IOException) {
+			assert(true)
+		}
+	}
+}

kotlin/src/nativeInterop/cinterop/libvosk.def

@@ -0,0 +1,25 @@
+headers = libvosk/vosk_api.h
+package = libvosk
+
+noStringConversion = \
+	vosk_batch_recognizer_accept_waveform \
+	vosk_recognizer_accept_waveform
+
+compilerOpts.linux = \
+	-I/usr/include/libvosk/ \
+	-I/usr/local/include/libvosk/ \
+	-I/usr/include/ \
+	-I/usr/local/include/
+
+compilerOpts.linux_x64 = \
+	-I/usr/lib64/libvosk/ \
+    -I/usr/local/lib64/libvosk/
+
+linkerOpts.linux = \
+	-L/usr/lib/ \
+	-L/usr/local/lib/ \
+	-llibvosk
+
+linkerOpts.linux_x64 = \
+	-L/usr/lib64/ \
+	-L/usr/local/lib64/

kotlin/src/nativeMain/kotlin/org/vosk/BatchModel.kt

@@ -0,0 +1,50 @@
+/*
+ * Copyright 2020 Alpha Cephei Inc. & Doomsdayrs
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import cnames.structs.VoskBatchModel
+import kotlinx.cinterop.CPointer
+import libvosk.vosk_batch_model_free
+import libvosk.vosk_batch_model_new
+import libvosk.vosk_batch_model_wait
+
+/**
+ * Batch model object
+ *
+ * 26 / 12 / 2022
+ */
+actual class BatchModel(val pointer: CPointer<VoskBatchModel>) : Freeable {
+	/**
+	 * Creates the batch recognizer object
+	 */
+	@Throws(IOException::class)
+	actual constructor(path: String) : this(vosk_batch_model_new(path) ?: throw ioException(path))
+
+	/**
+	 *  Releases batch model object
+	 */
+	actual override fun free() {
+		vosk_batch_model_free(pointer)
+	}
+
+	/**
+	 * Wait for the processing
+	 */
+	actual fun await() {
+		vosk_batch_model_wait(pointer)
+	}
+}

kotlin/src/nativeMain/kotlin/org/vosk/BatchRecognizer.kt

@@ -0,0 +1,86 @@
+/*
+ * Copyright 2020 Alpha Cephei Inc. & Doomsdayrs
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.vosk
+
+import cnames.structs.VoskBatchRecognizer
+import kotlinx.cinterop.CPointer
+import kotlinx.cinterop.toCValues
+import kotlinx.cinterop.toKString
+import libvosk.*
+
+/**
+ * Batch recognizer object
+ *
+ * 26 / 12 / 2022
+ */
+actual class BatchRecognizer(val pointer: CPointer<VoskBatchRecognizer>) : Freeable {
+	/**
+	 * Creates batch recognizer object
+	 */
+	actual constructor(
+		model: BatchModel,
+		sampleRate: Float
+	) : this(vosk_batch_recognizer_new(model.pointer, sampleRate)!!)
+
+	/**
+	 * Releases batch recognizer object
+	 */
+	actual override fun free() {
+		vosk_batch_recognizer_free(pointer)
+	}
+
+	/**
+	 * Accept batch voice data
+	 */
+	actual fun acceptWaveform(data: ByteArray) {
+		vosk_batch_recognizer_accept_waveform(pointer, data.toCValues(), data.size)
+	}
+
+	/**
+	 * Set NLSML output
+	 * @param nlsml - boolean value
+	 */
+	actual fun setNLSML(nlsml: Boolean) {
+		vosk_batch_recognizer_set_nlsml(pointer, nlsml.toInt())
+	}
+
+	/**
+	 *  Closes the stream
+	 */
+	actual fun finishStream() {
+		vosk_batch_recognizer_finish_stream(pointer)
+	}
+
+	/**
+	 * Return results
+	 */
+	actual val frontResult: String
+		get() = vosk_batch_recognizer_front_result(pointer)!!.toKString()
+
+	/**
+	 * Release and free first retrieved result
+	 */
+	actual fun pop() {
+		vosk_batch_recognizer_pop(pointer)
+	}
+
+	/**
+	 * Get amount of pending chunks for more intelligent waiting
+	 */
+	actual val pendingChunks: Int
+		get() = vosk_batch_recognizer_get_pending_chunks(pointer)
+}