Bump version to 0.3.32

are implemented, others on the way

.gitignore

@@ -1,20 +1,25 @@
-# Built application files
-*.apk
-*.ap_
+# Object files
+*.o
 
-# Java class files
-*.class
+# Temp files
+nohup.out
 
 # Gradle files
 .gradle/
-build/
 gradlew
 gradlew.bat
 gradle
-
-# Local configuration file (sdk path, etc)
 local.properties
 
+# Android
+android/build
+android/lib/build
+android/model-en/build
+android/model-en/src/main/assets/model-en-us
+android/repo
+*.apk
+*.ap_
+
 # Cmake
 .cxx
 cmake_install.cmake
@@ -24,12 +29,8 @@ wheelhouse
 __pycache__
 *.egg-info
 python/dist
-python/vosk/*.cc
-python/vosk/*.c
-python/vosk/*.h
-python/vosk/*.i
-python/vosk/vosk.py
-python/vosk/vosk_wrap.cpp
+python/build
+python/vosk/*.so
 python/test/db
 python/test/hyp
 python/test/model
@@ -38,23 +39,32 @@ python/test/result.txt
 python/test/wav.scp
 
 # Java
-*.so
-java/org
-java/*.cc
-java/model-spk/
-java/model/
+*.class
+java/lib/model
+java/demo/model
+java/lib/build
+java/demo/build
 
 # CSharp
-csharp/gen
-csharp/*.exe
-csharp/*.c
-csharp/model/
-csharp/test.wav
+*.dll
+*.so
+*.dylib
+*.nupkg
+csharp/demo/model
+csharp/demo/test.wav
+csharp/demo/bin
+csharp/demo/obj
 
 # Node
-nodejs/vosk_wrap.cc
-nodejs/example/model
-nodejs/example/test.wav
+nodejs/demo/model
+nodejs/demo/model-spk
+nodejs/demo/test.wav
 nodejs/node_modules
 nodejs/package-lock.json
-nodejs/build
+
+# C
+c/test_vosk
+c/test_vosk_speaker
+c/oprofile_data
+c/model
+c/test.wav

.travis.yml

@@ -7,11 +7,10 @@ matrix:
       services:
         - docker
       env: DOCKER_IMAGE=alphacep/kaldi-manylinux:latest
-           PLAT=manylinux2010_x86_64
 
 install:
   - docker pull $DOCKER_IMAGE
 
 script:
-  - docker run --rm -e PLAT=$PLAT -v `pwd`:/io $DOCKER_IMAGE $PRE_CMD /io/travis/build-wheels.sh
+  - docker run --rm -v `pwd`:/io $DOCKER_IMAGE $PRE_CMD /io/travis/build-wheels.sh
   - ls wheelhouse/

README.md

@@ -1,26 +1,26 @@
-Vosk is an open source speech recognition toolkit which supports 9
-languages - English, German, French, Spanish, Portuguese, Chinese,
-Russian, Turkish, Vietnamese. Vosk works offline with small (50 Mb), but
-accurate model, zero-latency response with streaming API, reconfigurable
+# About
+
+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 models are small (50 Mb) but provide continuous large vocabulary
+transcription, zero-latency response with streaming API, reconfigurable
 vocabulary and speaker identification.
 
-### Installation and usage
+Speech recognition bindings implemented for various programming languages
+like Python, Java, Node.JS, C#, C++ and others.
 
-For Vosk installation instructions, examples and turorial and documentation visit https://alphacephei.com/vosk
+Vosk supplies speech recognition for chatbots, smart home appliances,
+virtual assistants. It can also create subtitles for movies,
+transcription for lectures and interviews.
 
-### Build
+Vosk scales from small devices like Raspberry Pi or Android smartphone to
+big clusters.
 
-[![Build Status](https://travis-ci.com/alphacep/vosk-api.svg?branch=master)](https://travis-ci.com/alphacep/vosk-api)
+# Documentation
 
-### Models for different languages
-
-For information about models see [the documentation on available models](https://alphacephei.com/vosk/models.html).
-
-### Contact Us
-
-If you have any questions, feel free to:
-
-   * Post an issue here on github
-   * Send us an e-mail at [contact@alphacephei.com](mailto:contact@alphacephei.com)
-   * Join our group dedicated to speech recognition on Telegram [@speech_recognition](https://t.me/speech_recognition)
-   * We have a Wechat group which is pretty big, so it is invitation-only. Mail us to join the group and provide some information about yourself.
+For installation instructions, examples and documentation visit [Vosk
+Website](https://alphacephei.com/vosk).

android/CMakeLists.txt

@@ -1,67 +0,0 @@
-# Vosk CMake File
-
-cmake_minimum_required(VERSION 3.4.1)
-
-if ("x${ANDROID_ABI}" STREQUAL "xarmeabi-v7a")
-set(OPENBLAS_ARCH "armv7")
-set(KALDI_SUFFIX "arm_32")
-elseif ("x${ANDROID_ABI}" STREQUAL "xarm64-v8a")
-set(OPENBLAS_ARCH "armv8")
-set(KALDI_SUFFIX "arm_64")
-elseif ("x${ANDROID_ABI}" STREQUAL "xx86")
-set(OPENBLAS_ARCH "atom")
-set(KALDI_SUFFIX "x86")
-else ("x${ANDROID_ABI}" STREQUAL "xarmeabi-v7a")
-set(OPENBLAS_ARCH "atom")
-set(KALDI_SUFFIX "x86_64")
-endif ("x${ANDROID_ABI}" STREQUAL "xarmeabi-v7a")
-
-set(KALDI_ROOT "${PROJECT_SOURCE_DIR}/build/kaldi_${KALDI_SUFFIX}/kaldi")
-set(LIB_ROOT "${PROJECT_SOURCE_DIR}/build/kaldi_${KALDI_SUFFIX}/local")
-
-set(API_SOURCES
-"${PROJECT_SOURCE_DIR}/../src/kaldi_recognizer.cc"
-"${PROJECT_SOURCE_DIR}/../src/kaldi_recognizer.h"
-"${PROJECT_SOURCE_DIR}/../src/model.cc"
-"${PROJECT_SOURCE_DIR}/../src/model.h"
-"${PROJECT_SOURCE_DIR}/../src/spk_model.cc"
-"${PROJECT_SOURCE_DIR}/../src/spk_model.h"
-"${PROJECT_SOURCE_DIR}/../src/vosk_api.cc"
-"${PROJECT_SOURCE_DIR}/../src/vosk_api.h"
-)
-
-set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O3 -DFST_NO_DYNAMIC_LINKING")
-
-add_library( kaldi_jni SHARED
-             build/generated-src/cpp/vosk_wrap.cc
-             ${API_SOURCES}
-)
-
-include_directories("${PROJECT_SOURCE_DIR}/../src" "build/kaldi_${KALDI_SUFFIX}/kaldi/src" "build/kaldi_${KALDI_SUFFIX}/local/include")
-
-target_link_libraries( kaldi_jni
-             ${KALDI_ROOT}/src/online2/kaldi-online2.a
-             ${KALDI_ROOT}/src/decoder/kaldi-decoder.a
-             ${KALDI_ROOT}/src/ivector/kaldi-ivector.a
-             ${KALDI_ROOT}/src/gmm/kaldi-gmm.a
-             ${KALDI_ROOT}/src/nnet3/kaldi-nnet3.a
-             ${KALDI_ROOT}/src/tree/kaldi-tree.a
-             ${KALDI_ROOT}/src/feat/kaldi-feat.a
-             ${KALDI_ROOT}/src/lat/kaldi-lat.a
-             ${KALDI_ROOT}/src/lm/kaldi-lm.a
-             ${KALDI_ROOT}/src/hmm/kaldi-hmm.a
-             ${KALDI_ROOT}/src/transform/kaldi-transform.a
-             ${KALDI_ROOT}/src/cudamatrix/kaldi-cudamatrix.a
-             ${KALDI_ROOT}/src/matrix/kaldi-matrix.a
-             ${KALDI_ROOT}/src/fstext/kaldi-fstext.a
-             ${KALDI_ROOT}/src/util/kaldi-util.a
-             ${KALDI_ROOT}/src/base/kaldi-base.a
-             ${LIB_ROOT}/lib/libfst.a
-             ${LIB_ROOT}/lib/libfstngram.a
-             ${LIB_ROOT}/lib/libopenblas_${OPENBLAS_ARCH}-r0.3.7.a
-             ${LIB_ROOT}/lib/libclapack.a
-             ${LIB_ROOT}/lib/liblapack.a
-             ${LIB_ROOT}/lib/libblas.a
-             ${LIB_ROOT}/lib/libf2c.a
-             log
-             )

android/README.md

@@ -1,22 +1,3 @@
-This is still work in progress, more to come
+Vosk library for Android
 
-## TODO
-
-   * Optimize graph construction, current one is below accuracy
-
-   * Load model from the AAR (mmap them in tflite style)
-
-   * Add decoding speed measurement
-
-   * Add wakeup word
-
-   * Add speakerid
-
-   * Integrate proper hardware optimized neural network library. Candidates are:
-
-      * https://github.com/XiaoMi/mace
-      * https://github.com/Tencent/ncnn
-      * https://developer.android.com/ndk/guides/neuralnetworks/ (since API level 27)
-      * https://github.com/google/XNNPACK
-
-   * Quantization for the models
+See for details https://alphacephei.com/vosk/android

android/build.gradle

@@ -4,65 +4,55 @@ buildscript {
         jcenter()
     }
     dependencies {
-        classpath 'com.android.tools.build:gradle:3.5.3'
+        classpath 'com.android.tools.build:gradle:4.1.3'
     }
 }
 
 allprojects {
-  repositories {
-      google()
-      jcenter()
-  }
+    version = '0.3.30'
 }
 
-apply plugin: 'com.android.library'
+subprojects {
 
-android {
-    compileSdkVersion 29
-    defaultConfig {
-        minSdkVersion 21
-        targetSdkVersion 29
-        versionCode 5
-        versionName "5.2"
-        setProperty("archivesBaseName", "kaldi-android-$versionName")
-        externalNativeBuild {
-            cmake {
-                arguments "-DCMAKE_VERBOSE_MAKEFILE=ON", "-DANDROID_ARM_NEON=TRUE", "-DCMAKE_CXX_FLAGS_RELEASE=-O3"
+    apply plugin: 'com.android.library'
+    apply plugin: 'maven-publish'
+
+    repositories {
+        google()
+        jcenter()
+    }
+
+    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/'
+                    }
+                }
             }
         }
-        ndk {
-            abiFilters 'armeabi-v7a', 'arm64-v8a', 'x86_64', 'x86'
-        }
-    }
-    sourceSets {
-         main {
-              java.srcDirs = ['src/main/java', 'build/generated-src']
-         }
-    }
-    externalNativeBuild {
-       cmake {
-           path "CMakeLists.txt"
-       }
-    }
-}
-
-task swig {
-    doLast {
-        mkdir 'build/generated-src/java'
-        mkdir 'build/generated-src/cpp'
-        exec {
-            commandLine 'swig',
-            "-c++",
-            "-java", "-package", "org.kaldi",
-            "-outdir", "build/generated-src/java", "-o", "build/generated-src/cpp/vosk_wrap.cc",
-            "../src/vosk.i"
+        repositories {
+            maven {
+                url = "$rootDir/repo"
+            }
         }
     }
 }
-
-task kaldi(type: Exec) {
-    commandLine './build-kaldi.sh'
-    environment ANDROID_SDK_HOME: android.getSdkDirectory()
-}
-
-preBuild.dependsOn kaldi, swig

android/lib/build-vosk.sh

@@ -1,6 +1,6 @@
 #!/bin/bash
 
-# Copyright 2019 Alpha Cephei Inc.
+# Copyright 2019-2021 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.
@@ -14,70 +14,39 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-if [ "x$ANDROID_SDK_HOME" == "x" ]; then
-    echo "ANDROID_SDK_HOME environment variable is undefined, define it with local.properties or with export"
+if [ "x$ANDROID_NDK_HOME" == "x" ]; then
+    echo "ANDROID_NDK_HOME environment variable is undefined, define it with local.properties or with export"
     exit 1
 fi
 
-if [ ! -d "$ANDROID_SDK_HOME" ]; then
-    echo "ANDROID_SDK_HOME ($ANDROID_SDK_HOME) is missing. Make sure you have sdk installed"
-    exit 1
-fi
-
-if [ ! -d "$ANDROID_SDK_HOME/ndk-bundle" ]; then
-    echo "$ANDROID_SDK_HOME/ndk-bundle is missing. Make sure you have ndk installed within sdk"
+if [ ! -d "$ANDROID_NDK_HOME" ]; then
+    echo "ANDROID_NDK_HOME ($ANDROID_NDK_HOME) is missing. Make sure you have ndk installed"
     exit 1
 fi
 
 set -x
 
 OS_NAME=`echo $(uname -s) | tr '[:upper:]' '[:lower:]'`
-ANDROID_NDK_HOME=$ANDROID_SDK_HOME/ndk-bundle
 ANDROID_TOOLCHAIN_PATH=$ANDROID_NDK_HOME/toolchains/llvm/prebuilt/${OS_NAME}-x86_64
-WORKDIR_X86=`pwd`/build/kaldi_x86
-WORKDIR_X86_64=`pwd`/build/kaldi_x86_64
-WORKDIR_ARM32=`pwd`/build/kaldi_arm_32
-WORKDIR_ARM64=`pwd`/build/kaldi_arm_64
+WORKDIR_BASE=`pwd`/build
 PATH=$PATH:$ANDROID_NDK_HOME/toolchains/llvm/prebuilt/${OS_NAME}-x86_64/bin
-OPENFST_VERSION=1.6.7
+OPENFST_VERSION=1.8.0
 
-mkdir -p $WORKDIR_ARM64/local/lib $WORKDIR_ARM32/local/lib $WORKDIR_X86_64/local/lib $WORKDIR_X86/local/lib
+for arch in armeabi-v7a arm64-v8a x86_64 x86; do
 
-# Build standalone CLAPACK since gfortran is missing
-cd build
-git clone https://github.com/simonlynen/android_libs
-cd android_libs/lapack
-sed -i.bak -e 's/APP_STL := gnustl_static/APP_STL := c++_static/g' jni/Application.mk && \
-sed -i.bak -e 's/android-10/android-21/g' project.properties && \
-sed -i.bak -e 's/APP_ABI := armeabi armeabi-v7a/APP_ABI := armeabi-v7a arm64-v8a x86_64 x86/g' jni/Application.mk && \
-sed -i.bak -e 's/LOCAL_MODULE:= testlapack/#LOCAL_MODULE:= testlapack/g' jni/Android.mk && \
-sed -i.bak -e 's/LOCAL_SRC_FILES:= testclapack.cpp/#LOCAL_SRC_FILES:= testclapack.cpp/g' jni/Android.mk && \
-sed -i.bak -e 's/LOCAL_STATIC_LIBRARIES := lapack/#LOCAL_STATIC_LIBRARIES := lapack/g' jni/Android.mk && \
-sed -i.bak -e 's/include $(BUILD_SHARED_LIBRARY)/#include $(BUILD_SHARED_LIBRARY)/g' jni/Android.mk && \
-${ANDROID_NDK_HOME}/ndk-build && \
-cp obj/local/armeabi-v7a/*.a ${WORKDIR_ARM32}/local/lib && \
-cp obj/local/arm64-v8a/*.a ${WORKDIR_ARM64}/local/lib
-cp obj/local/x86_64/*.a ${WORKDIR_X86_64}/local/lib
-cp obj/local/x86/*.a ${WORKDIR_X86}/local/lib
-
-# Architecture-specific part
-
-for arch in arm32 arm64 x86_64 x86; do
-#for arch in x86_64; do
+WORKDIR=${WORKDIR_BASE}/kaldi_${arch}
 
 case $arch in
-    arm32)
+    armeabi-v7a)
           BLAS_ARCH=ARMV7
-          WORKDIR=$WORKDIR_ARM32
           HOST=arm-linux-androideabi
           AR=arm-linux-androideabi-ar
           CC=armv7a-linux-androideabi21-clang
           CXX=armv7a-linux-androideabi21-clang++
           ARCHFLAGS="-mfloat-abi=softfp -mfpu=neon"
           ;;
-    arm64)
+    arm64-v8a)
           BLAS_ARCH=ARMV8
-          WORKDIR=$WORKDIR_ARM64
           HOST=aarch64-linux-android
           AR=aarch64-linux-android-ar
           CC=aarch64-linux-android21-clang
@@ -86,7 +55,6 @@ case $arch in
           ;;
     x86_64)
           BLAS_ARCH=ATOM
-          WORKDIR=$WORKDIR_X86_64
           HOST=x86_64-linux-android
           AR=x86_64-linux-android-ar
           CC=x86_64-linux-android21-clang
@@ -95,7 +63,6 @@ case $arch in
           ;;
     x86)
           BLAS_ARCH=ATOM
-          WORKDIR=$WORKDIR_X86
           HOST=i686-linux-android
           AR=i686-linux-android-ar
           CC=i686-linux-android21-clang
@@ -104,12 +71,27 @@ case $arch in
           ;;
 esac
 
+mkdir -p $WORKDIR/local/lib
+
 # openblas first
 cd $WORKDIR
-git clone -b v0.3.7 --single-branch https://github.com/xianyi/OpenBLAS
+git clone -b v0.3.13 --single-branch https://github.com/xianyi/OpenBLAS
 make -C OpenBLAS TARGET=$BLAS_ARCH ONLY_CBLAS=1 AR=$AR CC=$CC HOSTCC=gcc ARM_SOFTFP_ABI=1 USE_THREAD=0 NUM_THREADS=1 -j4
 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 \
+    -DCMAKE_TRY_COMPILE_TARGET_TYPE=STATIC_LIBRARY \
+    -DCMAKE_CROSSCOMPILING=True ..
+make -j 8 -C F2CLIBS/libf2c
+make -j 8 -C BLAS/SRC
+make -j 8 -C SRC
+find . -name "*.a" | xargs cp -t $WORKDIR/local/lib
+
 # tools directory --> we'll only compile OpenFST
 cd $WORKDIR
 git clone https://github.com/alphacep/openfst
@@ -123,19 +105,25 @@ 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 $WORKDIR/kaldi/src
-if [ "`uname`" == "Darwin"  ]; then
-  sed -i.bak -e 's/libfst.dylib/libfst.a/' configure
-fi
-
 CXX=$CXX CXXFLAGS="$ARCHFLAGS -O3 -DFST_NO_DYNAMIC_LINKING" ./configure --use-cuda=no \
-    --mathlib=OPENBLAS --shared \
+    --mathlib=OPENBLAS_CLAPACK --shared \
     --android-incdir=${ANDROID_TOOLCHAIN_PATH}/sysroot/usr/include \
     --host=$HOST --openblas-root=${WORKDIR}/local \
     --fst-root=${WORKDIR}/local --fst-version=${OPENFST_VERSION}
-
 make -j 8 depend
-make -j 8 online2 lm
+cd $WORKDIR/kaldi/src
+make -j 8 online2 lm 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
 
 done

android/lib/build.gradle

@@ -0,0 +1,42 @@
+def archiveName = "vosk-android"
+def pomName = "Vosk Android"
+def pomDescription = "Vosk speech recognition library for Android"
+
+android {
+    compileSdkVersion 29
+    defaultConfig {
+        minSdkVersion 21
+        targetSdkVersion 29
+        versionCode 6
+        versionName = version
+        archivesBaseName = archiveName
+    }
+    compileOptions {
+        sourceCompatibility JavaVersion.VERSION_1_8
+        targetCompatibility JavaVersion.VERSION_1_8
+    }
+}
+
+task buildVosk(type: Exec) {
+    commandLine './build-vosk.sh'
+    environment ANDROID_NDK_HOME: android.getSdkDirectory()
+}
+
+dependencies {
+    implementation 'net.java.dev.jna:jna:4.4.0@aar'
+}
+
+//preBuild.dependsOn buildVosk
+
+publishing {
+    publications {
+        aar(MavenPublication) {
+            artifactId = archiveName
+            artifact("$buildDir/outputs/aar/$archiveName-release.aar")
+            pom {
+                name = pomName
+                description = pomDescription
+            }
+        }
+    }
+}

android/lib/settings.gradle

@@ -0,0 +1 @@
+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.kaldi">
+<manifest xmlns:android="http://schemas.android.com/apk/res/android" package="org.vosk">
 </manifest>

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

@@ -0,0 +1,60 @@
+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;
+import java.io.InputStream;
+import java.io.IOException;
+import java.nio.file.Files;
+import java.nio.file.StandardCopyOption;
+
+public class LibVosk {
+
+    static {
+        Native.register(LibVosk.class, "vosk");
+    }
+
+    public static native void vosk_set_log_level(int level);
+
+    public static native Pointer vosk_model_new(String path);
+
+    public static native void vosk_model_free(Pointer model);
+
+    public static native Pointer vosk_spk_model_new(String path);
+
+    public static native void vosk_spk_model_free(Pointer model);
+
+    public static native Pointer vosk_recognizer_new(Model model, float sample_rate);
+
+    public static native Pointer vosk_recognizer_new_spk(Pointer model, float sample_rate, Pointer spk_model);
+
+    public static native Pointer vosk_recognizer_new_grm(Pointer model, float sample_rate, String grammar);
+
+    public static native void vosk_recognizer_set_max_alternatives(Pointer recognizer, int max_alternatives);
+
+    public static native void vosk_recognizer_set_words(Pointer recognizer, boolean 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);
+
+    public static native boolean vosk_recognizer_accept_waveform_s(Pointer recognizer, short[] data, int len);
+
+    public static native boolean vosk_recognizer_accept_waveform_f(Pointer recognizer, float[] data, int len);
+
+    public static native String vosk_recognizer_result(Pointer recognizer);
+
+    public static native String vosk_recognizer_final_result(Pointer recognizer);
+
+    public static native String vosk_recognizer_partial_result(Pointer recognizer);
+
+    public static native void vosk_recognizer_reset(Pointer recognizer);
+
+    public static native void vosk_recognizer_free(Pointer recognizer);
+
+    public static void setLogLevel(LogLevel loglevel) {
+        vosk_set_log_level(loglevel.getValue());
+    }
+}

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

@@ -0,0 +1,17 @@
+package org.vosk;
+
+public enum LogLevel {
+    WARNINGS(-1),  // Print warning and errors
+    INFO(0),       // Print info, along with warning and error messages, but no debug
+    DEBUG(1);      // Print debug info
+
+    private final int value;
+
+    LogLevel(int value) {
+        this.value = value;
+    }
+
+    public int getValue() {
+        return this.value;
+    }
+}

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

@@ -0,0 +1,17 @@
+package org.vosk;
+
+import com.sun.jna.PointerType;
+
+public class Model extends PointerType implements AutoCloseable {
+    public Model() {
+    }
+
+    public Model(String path) {
+        super(LibVosk.vosk_model_new(path));
+    }
+
+    @Override
+    public void close() {
+        LibVosk.vosk_model_free(this.getPointer());
+    }
+}

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

@@ -0,0 +1,62 @@
+package org.vosk;
+
+import com.sun.jna.PointerType;
+
+public class Recognizer extends PointerType implements AutoCloseable {
+    public Recognizer(Model model, float sampleRate) {
+        super(LibVosk.vosk_recognizer_new(model, sampleRate));
+    }
+
+    public Recognizer(Model model, float sampleRate, SpeakerModel spkModel) {
+        super(LibVosk.vosk_recognizer_new_spk(model.getPointer(), sampleRate, spkModel.getPointer()));
+    }
+
+    public Recognizer(Model model, float sampleRate, String grammar) {
+        super(LibVosk.vosk_recognizer_new_grm(model.getPointer(), sampleRate, grammar));
+    }
+
+    public void setMaxAlternatives(int maxAlternatives) {
+        LibVosk.vosk_recognizer_set_max_alternatives(this.getPointer(), maxAlternatives);
+    }
+
+    public void setWords(boolean words) {
+        LibVosk.vosk_recognizer_set_words(this.getPointer(), words);
+    }
+
+    public void setSpeakerModel(SpeakerModel spkModel) {
+        LibVosk.vosk_recognizer_set_spk_model(this.getPointer(), spkModel.getPointer());
+    }
+
+    public boolean acceptWaveForm(byte[] data, int len) {
+        return LibVosk.vosk_recognizer_accept_waveform(this.getPointer(), data, len);
+    }
+
+    public boolean acceptWaveForm(short[] data, int len) {
+        return LibVosk.vosk_recognizer_accept_waveform_s(this.getPointer(), data, len);
+    }
+
+    public boolean acceptWaveForm(float[] data, int len) {
+        return LibVosk.vosk_recognizer_accept_waveform_f(this.getPointer(), data, len);
+    }
+
+    public String getResult() {
+        return LibVosk.vosk_recognizer_result(this.getPointer());
+    }
+
+    public String getPartialResult() {
+        return LibVosk.vosk_recognizer_partial_result(this.getPointer());
+    }
+
+    public String getFinalResult() {
+        return LibVosk.vosk_recognizer_final_result(this.getPointer());
+    }
+
+    public void reset() {
+        LibVosk.vosk_recognizer_reset(this.getPointer());
+    }
+
+    @Override
+    public void close() {
+        LibVosk.vosk_recognizer_free(this.getPointer());
+    }
+}

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

@@ -0,0 +1,17 @@
+package org.vosk;
+
+import com.sun.jna.PointerType;
+
+public class SpeakerModel extends PointerType implements AutoCloseable {
+    public SpeakerModel() {
+    }
+
+    public SpeakerModel(String path) {
+        super(LibVosk.vosk_spk_model_new(path));
+    }
+
+    @Override
+    public void close() {
+        LibVosk.vosk_spk_model_free(this.getPointer());
+    }
+}

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

@@ -12,28 +12,35 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-package org.kaldi;
+package org.vosk.android;
 
-/** Interface to receive recognition results */
+/**
+ * Interface to receive recognition results
+ */
 public interface RecognitionListener {
 
     /**
      * Called when partial recognition result is available.
      */
-    public void onPartialResult(String hypothesis);
+    void onPartialResult(String hypothesis);
 
     /**
-     * Called after the recognition is ended.
+     * Called after silence occured.
      */
-    public void onResult(String hypothesis);
+    void onResult(String hypothesis);
+
+    /**
+     * Called after stream end.
+     */
+    void onFinalResult(String hypothesis);
 
     /**
      * Called when an error occurs.
      */
-    public void onError(Exception exception);
+    void onError(Exception exception);
 
     /**
      * Called after timeout expired
      */
-    public void onTimeout();
+    void onTimeout();
 }

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

@@ -0,0 +1,257 @@
+// Copyright 2019 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.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;
+
+/**
+ * Service that records audio in a thread, passes it to a recognizer and emits
+ * recognition results. Recognition events are passed to a client using
+ * {@link RecognitionListener}
+ */
+public class SpeechService {
+
+    private final Recognizer recognizer;
+
+    private final int sampleRate;
+    private final static float BUFFER_SIZE_SECONDS = 0.2f;
+    private final int bufferSize;
+    private final AudioRecord recorder;
+
+    private RecognizerThread recognizerThread;
+
+    private final Handler mainHandler = new Handler(Looper.getMainLooper());
+
+    /**
+     * Creates speech service. Service holds the AudioRecord object, so you
+     * need to call {@link #shutdown()} in order to properly finalize it.
+     *
+     * @throws IOException thrown if audio recorder can not be created for some reason.
+     */
+    public SpeechService(Recognizer recognizer, float sampleRate) throws IOException {
+        this.recognizer = recognizer;
+        this.sampleRate = (int) sampleRate;
+
+        bufferSize = Math.round(this.sampleRate * BUFFER_SIZE_SECONDS);
+        recorder = new AudioRecord(
+                AudioSource.VOICE_RECOGNITION, this.sampleRate,
+                AudioFormat.CHANNEL_IN_MONO,
+                AudioFormat.ENCODING_PCM_16BIT, bufferSize * 2);
+
+        if (recorder.getState() == AudioRecord.STATE_UNINITIALIZED) {
+            recorder.release();
+            throw new 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
+     */
+    public boolean startListening(RecognitionListener listener) {
+        if (null != recognizerThread)
+            return false;
+
+        recognizerThread = new 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.
+     * <p>
+     * timeout - timeout in milliseconds to listen.
+     *
+     * @return true if recognition was actually started
+     */
+    public boolean startListening(RecognitionListener listener, int timeout) {
+        if (null != recognizerThread)
+            return false;
+
+        recognizerThread = new RecognizerThread(listener, timeout);
+        recognizerThread.start();
+        return true;
+    }
+
+    private boolean stopRecognizerThread() {
+        if (null == recognizerThread)
+            return false;
+
+        try {
+            recognizerThread.interrupt();
+            recognizerThread.join();
+        } catch (InterruptedException e) {
+            // 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
+     */
+    public boolean stop() {
+        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
+     */
+    public boolean cancel() {
+        if (recognizerThread != null) {
+            recognizerThread.setPause(true);
+        }
+        return stopRecognizerThread();
+    }
+
+    /**
+     * Shutdown the recognizer and release the recorder
+     */
+    public void shutdown() {
+        recorder.release();
+    }
+
+    public void setPause(boolean paused) {
+        if (recognizerThread != null) {
+            recognizerThread.setPause(paused);
+        }
+    }
+
+    /**
+     * Resets recognizer in a thread, starts recognition over again
+     */
+    public void reset() {
+        if (recognizerThread != null) {
+            recognizerThread.reset();
+        }
+    }
+
+    private final class RecognizerThread extends Thread {
+
+        private int remainingSamples;
+        private final int timeoutSamples;
+        private final static int NO_TIMEOUT = -1;
+        private volatile boolean paused = false;
+        private volatile boolean reset = false;
+
+        RecognitionListener listener;
+
+        public RecognizerThread(RecognitionListener listener, int timeout) {
+            this.listener = listener;
+            if (timeout != NO_TIMEOUT)
+                this.timeoutSamples = timeout * sampleRate / 1000;
+            else
+                this.timeoutSamples = NO_TIMEOUT;
+            this.remainingSamples = this.timeoutSamples;
+        }
+
+        public RecognizerThread(RecognitionListener listener) {
+            this(listener, NO_TIMEOUT);
+        }
+
+        /**
+         * When we are paused, don't process audio by the recognizer and don't emit
+         * any listener results
+         *
+         * @param paused the status of pause
+         */
+        public void setPause(boolean paused) {
+            this.paused = paused;
+        }
+
+        /**
+         * Set reset state to signal reset of the recognizer and start over
+         */
+        public void reset() {
+            this.reset = true;
+        }
+
+        @Override
+        public void run() {
+
+            recorder.startRecording();
+            if (recorder.getRecordingState() == AudioRecord.RECORDSTATE_STOPPED) {
+                recorder.stop();
+                IOException ioe = new IOException(
+                        "Failed to start recording. Microphone might be already in use.");
+                mainHandler.post(() -> listener.onError(ioe));
+            }
+
+            short[] buffer = new short[bufferSize];
+
+            while (!interrupted()
+                    && ((timeoutSamples == NO_TIMEOUT) || (remainingSamples > 0))) {
+                int nread = recorder.read(buffer, 0, buffer.length);
+
+                if (paused) {
+                    continue;
+                }
+
+                if (reset) {
+                    recognizer.reset();
+                    reset = false;
+                }
+
+                if (nread < 0)
+                    throw new RuntimeException("error reading audio buffer");
+
+                if (recognizer.acceptWaveForm(buffer, nread)) {
+                    final String result = recognizer.getResult();
+                    mainHandler.post(() -> listener.onResult(result));
+                } else {
+                    final String partialResult = recognizer.getPartialResult();
+                    mainHandler.post(() -> listener.onPartialResult(partialResult));
+                }
+
+                if (timeoutSamples != NO_TIMEOUT) {
+                    remainingSamples = 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 {
+                    final String finalResult = recognizer.getFinalResult();
+                    mainHandler.post(() -> listener.onFinalResult(finalResult));
+                }
+            }
+
+        }
+    }
+}

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

@@ -0,0 +1,165 @@
+// Copyright 2019 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;
+
+/**
+ * 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
+ * {@link RecognitionListener}
+ */
+public class SpeechStreamService {
+
+    private final Recognizer recognizer;
+    private final InputStream inputStream;
+    private final int sampleRate;
+    private final static float BUFFER_SIZE_SECONDS = 0.2f;
+    private final int bufferSize;
+
+    private Thread recognizerThread;
+
+    private final Handler mainHandler = new Handler(Looper.getMainLooper());
+
+    /**
+     * Creates speech service.
+     **/
+    public SpeechStreamService(Recognizer recognizer, InputStream inputStream, float sampleRate) {
+        this.recognizer = recognizer;
+        this.sampleRate = (int) sampleRate;
+        this.inputStream = inputStream;
+        bufferSize = Math.round(this.sampleRate * BUFFER_SIZE_SECONDS * 2);
+    }
+
+    /**
+     * Starts recognition. Does nothing if recognition is active.
+     *
+     * @return true if recognition was actually started
+     */
+    public boolean start(RecognitionListener listener) {
+        if (null != recognizerThread)
+            return false;
+
+        recognizerThread = new 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.
+     * <p>
+     * timeout - timeout in milliseconds to listen.
+     *
+     * @return true if recognition was actually started
+     */
+    public boolean start(RecognitionListener listener, int timeout) {
+        if (null != recognizerThread)
+            return false;
+
+        recognizerThread = new 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
+     */
+    public boolean stop() {
+        if (null == recognizerThread)
+            return false;
+
+        try {
+            recognizerThread.interrupt();
+            recognizerThread.join();
+        } catch (InterruptedException e) {
+            // Restore the interrupted status.
+            Thread.currentThread().interrupt();
+        }
+
+        recognizerThread = null;
+        return true;
+    }
+
+    private final class RecognizerThread extends Thread {
+
+        private int remainingSamples;
+        private final int timeoutSamples;
+        private final static int NO_TIMEOUT = -1;
+        RecognitionListener listener;
+
+        public RecognizerThread(RecognitionListener listener, int timeout) {
+            this.listener = listener;
+            if (timeout != NO_TIMEOUT)
+                this.timeoutSamples = timeout * sampleRate / 1000;
+            else
+                this.timeoutSamples = NO_TIMEOUT;
+            this.remainingSamples = this.timeoutSamples;
+        }
+
+        public RecognizerThread(RecognitionListener listener) {
+            this(listener, NO_TIMEOUT);
+        }
+
+        @Override
+        public void run() {
+
+            byte[] buffer = new byte[bufferSize];
+
+            while (!interrupted()
+                    && ((timeoutSamples == NO_TIMEOUT) || (remainingSamples > 0))) {
+                try {
+                    int nread = inputStream.read(buffer, 0, buffer.length);
+                    if (nread < 0) {
+                        break;
+                    } else {
+                        boolean isSilence = recognizer.acceptWaveForm(buffer, nread);
+                        if (isSilence) {
+                            final String result = recognizer.getResult();
+                            mainHandler.post(() -> listener.onResult(result));
+                        } else {
+                            final String partialResult = recognizer.getPartialResult();
+                            mainHandler.post(() -> listener.onPartialResult(partialResult));
+                        }
+                    }
+
+                    if (timeoutSamples != NO_TIMEOUT) {
+                        remainingSamples = remainingSamples - nread;
+                    }
+
+                } catch (IOException e) {
+                    mainHandler.post(() -> listener.onError(e));
+                }
+            }
+
+            // If we met timeout signal that speech ended
+            if (timeoutSamples != NO_TIMEOUT && remainingSamples <= 0) {
+                mainHandler.post(() -> listener.onTimeout());
+            } else {
+                final String finalResult = recognizer.getFinalResult();
+                mainHandler.post(() -> listener.onFinalResult(finalResult));
+            }
+        }
+    }
+}

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

@@ -0,0 +1,150 @@
+// Copyright 2019 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.BufferedReader;
+import java.io.File;
+import java.io.FileInputStream;
+import java.io.FileNotFoundException;
+import java.io.FileOutputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.io.InputStreamReader;
+import java.io.OutputStream;
+import java.util.concurrent.Executor;
+import java.util.concurrent.Executors;
+
+/**
+ * Provides utility methods to sync model files to external storage to allow
+ * C++ code access them. Relies on file named "uuid" to track updates.
+ */
+public class StorageService {
+
+    protected static final String TAG = StorageService.class.getSimpleName();
+
+    public interface Callback<R> {
+        void onComplete(R result);
+    }
+
+    public static void unpack(Context context, String sourcePath, final String targetPath, final Callback<Model> completeCallback, final Callback<IOException> errorCallback) {
+        Executor executor = Executors.newSingleThreadExecutor(); // change according to your requirements
+        Handler handler = new Handler(Looper.getMainLooper());
+        executor.execute(() -> {
+            try {
+                final String outputPath = sync(context, sourcePath, targetPath);
+                Model model = new Model(outputPath);
+                handler.post(() -> completeCallback.onComplete(model));
+            } catch (final IOException e) {
+                handler.post(() -> errorCallback.onComplete(e));
+            }
+        });
+    }
+
+    public static String sync(Context context, String sourcePath, String targetPath) throws IOException {
+
+        AssetManager assetManager = context.getAssets();
+
+        File externalFilesDir = context.getExternalFilesDir(null);
+        if (externalFilesDir == null) {
+            throw new IOException("cannot get external files dir, "
+                    + "external storage state is " + Environment.getExternalStorageState());
+        }
+
+        File targetDir = new File(externalFilesDir, targetPath);
+        String resultPath = new File(targetDir, sourcePath).getAbsolutePath();
+        String sourceUUID = readLine(assetManager.open(sourcePath + "/uuid"));
+        try {
+            String targetUUID = readLine(new FileInputStream(new File(targetDir, sourcePath + "/uuid")));
+            if (targetUUID.equals(sourceUUID)) return resultPath;
+        } catch (FileNotFoundException e) {
+            // ignore
+        }
+        deleteContents(targetDir);
+
+        copyAssets(assetManager, sourcePath, targetDir);
+
+        // Copy uuid
+        copyFile(assetManager, sourcePath + "/uuid", targetDir);
+
+        return resultPath;
+    }
+
+    private static String readLine(InputStream is) throws IOException {
+        return new BufferedReader(new InputStreamReader(is)).readLine();
+    }
+
+    private static boolean deleteContents(File dir) {
+        File[] files = dir.listFiles();
+        boolean success = true;
+        if (files != null) {
+            for (File file : files) {
+                if (file.isDirectory()) {
+                    success &= deleteContents(file);
+                }
+                if (!file.delete()) {
+                    success = false;
+                }
+            }
+        }
+        return success;
+    }
+
+    private static void copyAssets(AssetManager assetManager, String path, File outPath) throws IOException {
+        String[] assets = assetManager.list(path);
+        if (assets == null) {
+            return;
+        }
+        if (assets.length == 0) {
+            if (!path.endsWith("uuid"))
+                copyFile(assetManager, path, outPath);
+        } else {
+            File dir = new File(outPath, path);
+            if (!dir.exists()) {
+                Log.v(TAG, "Making directory " + dir.getAbsolutePath());
+                if (!dir.mkdirs()) {
+                    Log.v(TAG, "Failed to create directory " + dir.getAbsolutePath());
+                }
+            }
+            for (String asset : assets) {
+                copyAssets(assetManager, path + "/" + asset, outPath);
+            }
+        }
+    }
+
+    private static void copyFile(AssetManager assetManager, String fileName, File outPath) throws IOException {
+        InputStream in;
+
+        Log.v(TAG, "Copy " + fileName + " to " + outPath);
+        in = assetManager.open(fileName);
+        OutputStream out = new FileOutputStream(outPath + "/" + fileName);
+
+        byte[] buffer = new byte[4000];
+        int read;
+        while ((read = in.read(buffer)) != -1) {
+            out.write(buffer, 0, read);
+        }
+        in.close();
+        out.close();
+    }
+}

android/model-en/build.gradle

@@ -0,0 +1,47 @@
+def archiveName = "vosk-model-en"
+def pomName = "Vosk English Model"
+def pomDescription = "Small English model for Android"
+
+android {
+    compileSdkVersion 29
+    defaultConfig {
+        minSdkVersion 21
+        targetSdkVersion 29
+        versionCode 6
+        versionName = version
+        archivesBaseName = archiveName
+    }
+    buildFeatures {
+        buildConfig = false
+    }
+    sourceSets {
+        main {
+            assets.srcDirs += "$buildDir/generated/assets"
+        }
+    }
+}
+
+tasks.register('genUUID') {
+    def uuid = UUID.randomUUID().toString()
+    def odir = file("$buildDir/generated/assets/model-en-us")
+    def ofile = file("$odir/uuid")
+    doLast {
+        mkdir odir
+        ofile.text = uuid
+    }
+}
+
+preBuild.dependsOn(genUUID)
+
+publishing {
+    publications {
+        aar(MavenPublication) {
+            artifactId = archiveName
+            artifact("$buildDir/outputs/aar/$archiveName-release.aar")
+            pom {
+                name = pomName
+                description = pomDescription
+            }
+        }
+    }
+}

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

@@ -0,0 +1,3 @@
+<manifest xmlns:android="http://schemas.android.com/apk/res/android"
+    package="org.vosk.model.en">
+</manifest>

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

@@ -0,0 +1,7 @@
+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)

android/settings.gradle

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

android/src/main/java/org/kaldi/Assets.java

@@ -1,262 +0,0 @@
-// Copyright 2019 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.kaldi;
-
-import java.io.BufferedReader;
-import java.io.File;
-import java.io.FileInputStream;
-import java.io.FileOutputStream;
-import java.io.IOException;
-import java.io.InputStream;
-import java.io.InputStreamReader;
-import java.io.OutputStream;
-import java.io.PrintWriter;
-import java.io.Reader;
-import java.util.ArrayDeque;
-import java.util.ArrayList;
-import java.util.Collection;
-import java.util.Collections;
-import java.util.HashMap;
-import java.util.List;
-import java.util.Map;
-import java.util.Queue;
-
-import android.content.Context;
-import android.content.res.AssetManager;
-import android.os.Environment;
-import android.util.Log;
-
-/**
- * Provides utility methods to keep asset files to external storage to allow
- * further JNI code access assets from a filesystem.
- * 
- * There must be special file {@value #ASSET_LIST_NAME} among the application
- * assets containing relative paths of assets to synchronize. If the
- * corresponding path does not exist on the external storage it is copied. If
- * the path exists checksums are compared and the asset is copied only if there
- * is a mismatch. Checksum is stored in a separate asset with the name that
- * consists of the original name and a suffix that depends on the checksum
- * algorithm (e.g. MD5). Checksum files are copied along with the corresponding
- * asset files.
- * 
- * @author Alexander Solovets
- */
-public class Assets {
-
-    protected static final String TAG = Assets.class.getSimpleName();
-
-    public static final String ASSET_LIST_NAME = "assets.lst";
-    public static final String SYNC_DIR = "sync";
-    public static final String HASH_EXT = ".md5";
-
-    private final AssetManager assetManager;
-    private final File externalDir;
-
-    /**
-     * Creates new instance for asset synchronization
-     * 
-     * @param context
-     *            application context
-     * 
-     * @throws IOException
-     *             if the directory does not exist
-     * 
-     * @see android.content.Context#getExternalFilesDir
-     * @see android.os.Environment#getExternalStorageState
-     */
-    public Assets(Context context) throws IOException {
-        File appDir = context.getExternalFilesDir(null);
-        if (null == appDir)
-            throw new IOException("cannot get external files dir, "
-                    + "external storage state is " + Environment.getExternalStorageState());
-        externalDir = new File(appDir, SYNC_DIR);
-        assetManager = context.getAssets();
-    }
-
-    /**
-     * Creates new instance with specified destination for assets
-     * 
-     * @param context
-     *            application context to retrieve the assets
-     * @param path
-     *            path to sync the files
-     */ 
-    public Assets(Context context, String dest) {
-        externalDir = new File(dest);
-        assetManager = context.getAssets();
-    }
-
-    /**
-     * Returns destination path on external storage where assets are copied.
-     * 
-     * @return path to application directory or null if it does not exists
-     */
-    public File getExternalDir() {
-        return externalDir;
-    }
-
-    /**
-     * Returns the map of asset paths to the files checksums.
-     * 
-     * @return path to the root of resources directory on external storage
-     * @throws IOException
-     *             if an I/O error occurs or "assets.lst" is missing
-     */
-    public Map<String, String> getItems() throws IOException {
-        Map<String, String> items = new HashMap<String, String>();
-        for (String path : readLines(openAsset(ASSET_LIST_NAME))) {
-            Reader reader = new InputStreamReader(openAsset(path + HASH_EXT));
-            items.put(path, new BufferedReader(reader).readLine());
-        }
-        return items;
-    }
-
-    /**
-     * Returns path to hash mappings for the previously copied files. This
-     * method can be used to find out assets which must be updated.
-     */
-    public Map<String, String> getExternalItems() {
-        try {
-            Map<String, String> items = new HashMap<String, String>();
-            File assetFile = new File(externalDir, ASSET_LIST_NAME);
-            for (String line : readLines(new FileInputStream(assetFile))) {
-                String[] fields = line.split(" ");
-                items.put(fields[0], fields[1]);
-            }
-            return items;
-        } catch (IOException e) {
-            return Collections.emptyMap();
-        }
-    }
-
-    /**
-     * In case you want to create more smart sync implementation, this method
-     * returns the list of items which must be synchronized.
-     */
-    public Collection<String> getItemsToCopy(String path) throws IOException {
-        Collection<String> items = new ArrayList<String>();
-        Queue<String> queue = new ArrayDeque<String>();
-        queue.offer(path);
-
-        while (!queue.isEmpty()) {
-            path = queue.poll();
-            String[] list = assetManager.list(path);
-            for (String nested : list)
-                queue.offer(nested);
-
-            if (list.length == 0)
-                items.add(path);
-        }
-
-        return items;
-    }
-
-    private List<String> readLines(InputStream source) throws IOException {
-        List<String> lines = new ArrayList<String>();
-        BufferedReader br = new BufferedReader(new InputStreamReader(source));
-        String line;
-        while (null != (line = br.readLine()))
-            lines.add(line);
-        return lines;
-    }
-
-    private InputStream openAsset(String asset) throws IOException {
-        return assetManager.open(new File(SYNC_DIR, asset).getPath());
-    }
-
-    /**
-     * Saves the list of synchronized items. The list is stored as a two-column
-     * space-separated list of items in a text file. The file is located at the
-     * root of synchronization directory in the external storage.
-     * 
-     * @param items
-     *            the items
-     * @throws IOException
-     *             if an I/O error occurs
-     */
-    public void updateItemList(Map<String, String> items) throws IOException {
-        File assetListFile = new File(externalDir, ASSET_LIST_NAME);
-        PrintWriter pw = new PrintWriter(new FileOutputStream(assetListFile));
-        for (Map.Entry<String, String> entry : items.entrySet())
-            pw.format("%s %s\n", entry.getKey(), entry.getValue());
-        pw.close();
-    }
-
-    /**
-     * Copies raw asset resource to external storage of the device.
-     * 
-     * @param path
-     *            path of the asset to copy
-     * @throws IOException
-     *             if an I/O error occurs
-     */
-    public File copy(String asset) throws IOException {
-        InputStream source = openAsset(asset);
-        File destinationFile = new File(externalDir, asset);
-        destinationFile.getParentFile().mkdirs();
-        OutputStream destination = new FileOutputStream(destinationFile);
-        byte[] buffer = new byte[1024];
-        int nread;
-
-        while ((nread = source.read(buffer)) != -1) {
-            if (nread == 0) {
-                nread = source.read();
-                if (nread < 0)
-                    break;
-                destination.write(nread);
-                continue;
-            }
-            destination.write(buffer, 0, nread);
-        }
-        destination.close();
-        return destinationFile;
-    }
-
-    /**
-     * Performs the sync of assets in the application and on the external
-     * storage
-     * 
-     * @return The folder on external storage with data
-     * @throws IOException
-     */
-    public File syncAssets() throws IOException {
-        Collection<String> newItems = new ArrayList<String>();
-        Collection<String> unusedItems = new ArrayList<String>();
-        Map<String, String> items = getItems();
-        Map<String, String> externalItems = getExternalItems();
-
-        for (String path : items.keySet()) {
-            if (!items.get(path).equals(externalItems.get(path))
-                    || !(new File(externalDir, path).exists()))
-                newItems.add(path);
-        }
-
-        unusedItems.addAll(externalItems.keySet());
-        unusedItems.removeAll(items.keySet());
-
-        for (String path : newItems) {
-            File file = copy(path);
-        }
-
-        for (String path : unusedItems) {
-            File file = new File(externalDir, path);
-            file.delete();
-        }
-
-        updateItemList(items);
-        return externalDir;
-    }
-
-}

android/src/main/java/org/kaldi/SpeechRecognizer.java

@@ -1,302 +0,0 @@
-// Copyright 2019 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.kaldi;
-
-import static java.lang.String.format;
-
-import java.io.File;
-import java.io.IOException;
-import java.util.Collection;
-import java.util.HashSet;
-
-import android.media.AudioFormat;
-import android.media.AudioRecord;
-import android.media.MediaRecorder.AudioSource;
-import android.os.Handler;
-import android.os.Looper;
-import android.util.Log;
-
-/**
- * Main class to access recognizer functions. After configuration this class
- * starts a listener thread which records the data and recognizes it using
- * VOSK engine. Recognition events are passed to a client using
- * {@link RecognitionListener}
- * 
- */
-public class SpeechRecognizer {
-
-    protected static final String TAG = SpeechRecognizer.class.getSimpleName();
-
-    private final KaldiRecognizer recognizer;
-
-    private final int sampleRate;        
-    private final static float BUFFER_SIZE_SECONDS = 0.4f;
-    private int bufferSize;
-    private final AudioRecord recorder;
-    
-    private Thread recognizerThread;
-
-    private final Handler mainHandler = new Handler(Looper.getMainLooper());
-    
-    private final Collection<RecognitionListener> listeners = new HashSet<RecognitionListener>();
-    
-    /**
-     * Creates speech recognizer. Recognizer holds the AudioRecord object, so you 
-     * need to call {@link release} in order to properly finalize it.
-     * 
-     * @throws IOException thrown if audio recorder can not be created for some reason.
-     */
-    public SpeechRecognizer(Model model) throws IOException {
-        recognizer = new KaldiRecognizer(model, 16000.0f);
-        sampleRate = 16000;
-        bufferSize = Math.round(sampleRate * BUFFER_SIZE_SECONDS);
-        recorder = new AudioRecord(
-                AudioSource.VOICE_RECOGNITION, sampleRate,
-                AudioFormat.CHANNEL_IN_MONO,
-                AudioFormat.ENCODING_PCM_16BIT, bufferSize * 2);
-
-        if (recorder.getState() == AudioRecord.STATE_UNINITIALIZED) {
-            recorder.release();
-            throw new IOException(
-                    "Failed to initialize recorder. Microphone might be already in use.");
-        }
-    }
-
-    public SpeechRecognizer(Model model, SpkModel spkModel) throws IOException {
-        recognizer = new KaldiRecognizer(model, spkModel, 16000.0f);
-        sampleRate = 16000;
-        bufferSize = Math.round(sampleRate * BUFFER_SIZE_SECONDS);
-        recorder = new AudioRecord(
-                AudioSource.VOICE_RECOGNITION, sampleRate,
-                AudioFormat.CHANNEL_IN_MONO,
-                AudioFormat.ENCODING_PCM_16BIT, bufferSize * 2);
-
-        if (recorder.getState() == AudioRecord.STATE_UNINITIALIZED) {
-            recorder.release();
-            throw new IOException(
-                    "Failed to initialize recorder. Microphone might be already in use.");
-        }
-    }
-
-    /**
-     * Adds listener.
-     */
-    public void addListener(RecognitionListener listener) {
-        synchronized (listeners) {
-            listeners.add(listener);
-        }
-    }
-
-    /**
-     * Removes listener.
-     */
-    public void removeListener(RecognitionListener listener) {
-        synchronized (listeners) {
-            listeners.remove(listener);
-        }
-    }
-
-    /**
-     * Starts recognition. Does nothing if recognition is active.
-     * 
-     * @return true if recognition was actually started
-     */
-    public boolean startListening() {
-        if (null != recognizerThread)
-            return false;
-
-        recognizerThread = new RecognizerThread();
-        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
-     */
-    public boolean startListening(int timeout) {
-        if (null != recognizerThread)
-            return false;
-
-        recognizerThread = new RecognizerThread(timeout);
-        recognizerThread.start();
-        return true;
-    }
-
-    private boolean stopRecognizerThread() {
-        if (null == recognizerThread)
-            return false;
-
-        try {
-            recognizerThread.interrupt();
-            recognizerThread.join();
-        } catch (InterruptedException e) {
-            // Restore the interrupted status.
-            Thread.currentThread().interrupt();
-        }
-
-        recognizerThread = null;
-        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
-     */
-    public boolean stop() {
-        boolean result = stopRecognizerThread();
-        if (result) {
-            mainHandler.post(new ResultEvent(recognizer.Result(), true));
-        }
-        return result;
-    }
-
-    /**
-     * Cancels recognition. Listeners do not receive final result. Does nothing
-     * if recognition is not active.
-     * 
-     * @return true if recognition was actually canceled
-     */
-    public boolean cancel() {
-        boolean result = stopRecognizerThread();
-        recognizer.Result(); // Reset recognizer state
-        return result;
-    }
-    
-    /**
-     * Shutdown the recognizer and release the recorder
-     */
-    public void shutdown() {
-        recorder.release();
-    }
-    
-    private final class RecognizerThread extends Thread {
-        
-        private int remainingSamples;
-        private int timeoutSamples;
-        private final static int NO_TIMEOUT = -1;
-
-        public RecognizerThread(int timeout) {
-            if (timeout != NO_TIMEOUT)
-                this.timeoutSamples = timeout * sampleRate / 1000;
-            else
-                this.timeoutSamples = NO_TIMEOUT;
-            this.remainingSamples = this.timeoutSamples;
-        }
-
-        public RecognizerThread() {
-            this(NO_TIMEOUT);
-        }
-
-        @Override
-        public void run() {
-
-            recorder.startRecording();
-            if (recorder.getRecordingState() == AudioRecord.RECORDSTATE_STOPPED) {
-                recorder.stop();
-                IOException ioe = new IOException(
-                        "Failed to start recording. Microphone might be already in use.");
-                mainHandler.post(new OnErrorEvent(ioe));
-                return;
-            }
-
-            short[] buffer = new short[bufferSize];
-
-            while (!interrupted()
-                    && ((timeoutSamples == NO_TIMEOUT) || (remainingSamples > 0))) {
-                int nread = recorder.read(buffer, 0, buffer.length);
-
-                if (nread < 0) {
-                    throw new RuntimeException("error reading audio buffer");
-                } else {
-                    boolean isFinal = recognizer.AcceptWaveform(buffer, nread);
-                    if (isFinal) {
-                        mainHandler.post(new ResultEvent(recognizer.Result(), true));
-                    } else {
-                        mainHandler.post(new ResultEvent(recognizer.PartialResult(), false));
-                    }
-                }
-
-                if (timeoutSamples != NO_TIMEOUT) {
-                    remainingSamples = remainingSamples - nread;
-                }
-            }
-
-            recorder.stop();
-
-            // Remove all pending notifications.
-            mainHandler.removeCallbacksAndMessages(null);
-
-            // If we met timeout signal that speech ended
-            if (timeoutSamples != NO_TIMEOUT && remainingSamples <= 0) {
-                mainHandler.post(new TimeoutEvent());
-            }
-        }
-    }
-
-    private abstract class RecognitionEvent implements Runnable {
-        public void run() {
-            RecognitionListener[] emptyArray = new RecognitionListener[0];
-            for (RecognitionListener listener : listeners.toArray(emptyArray))
-                execute(listener);
-        }
-
-        protected abstract void execute(RecognitionListener listener);
-    }
-
-    private class ResultEvent extends RecognitionEvent {
-        protected final String hypothesis;
-        private final boolean finalResult;
-
-        ResultEvent(String hypothesis, boolean finalResult) {
-            this.hypothesis = hypothesis;
-            this.finalResult = finalResult;
-        }
-
-        @Override
-        protected void execute(RecognitionListener listener) {
-            if (finalResult)
-                listener.onResult(hypothesis);
-            else
-                listener.onPartialResult(hypothesis);
-        }
-    }
-
-    private class OnErrorEvent extends RecognitionEvent {
-        private final Exception exception;
-
-        OnErrorEvent(Exception exception) {
-            this.exception = exception;
-        }
-
-        @Override
-        protected void execute(RecognitionListener listener) {
-            listener.onError(exception);
-        }
-    }
-
-    private class TimeoutEvent extends RecognitionEvent {
-        @Override
-        protected void execute(RecognitionListener listener) {
-            listener.onTimeout();
-        }
-    }
-}

c/Makefile

@@ -0,0 +1,16 @@
+CFLAGS=-I../src
+LDFLAGS=-L../src -lvosk -ldl -lpthread -Wl,-rpath=../src
+
+all: test_vosk test_vosk_speaker
+
+test_vosk: test_vosk.o
+	g++ $^ -o $@ $(LDFLAGS)
+
+test_vosk_speaker: test_vosk_speaker.o
+	g++ $^ -o $@ $(LDFLAGS)
+
+%.o: %.c
+	g++ $(CFLAGS) -c -o $@ $<
+
+clean:
+	rm -f *.o *.a test_vosk test_vosk_speaker

c/test_vosk.c

@@ -0,0 +1,29 @@
+#include <vosk_api.h>
+#include <stdio.h>
+
+int main() {
+    FILE *wavin;
+    char buf[3200];
+    int nread, final;
+
+    VoskModel *model = vosk_model_new("model");
+    VoskRecognizer *recognizer = vosk_recognizer_new(model, 16000.0);
+
+    wavin = fopen("test.wav", "rb");
+    fseek(wavin, 44, SEEK_SET);
+    while (!feof(wavin)) {
+         nread = fread(buf, 1, sizeof(buf), wavin);
+         final = vosk_recognizer_accept_waveform(recognizer, buf, nread);
+         if (final) {
+             printf("%s\n", vosk_recognizer_result(recognizer));
+         } else {
+             printf("%s\n", vosk_recognizer_partial_result(recognizer));
+         }
+    }
+    printf("%s\n", vosk_recognizer_final_result(recognizer));
+
+    vosk_recognizer_free(recognizer);
+    vosk_model_free(model);
+    fclose(wavin);
+    return 0;
+}

c/test_vosk_speaker.c

@@ -0,0 +1,30 @@
+#include <vosk_api.h>
+#include <stdio.h>
+
+int main() {
+    FILE *wavin;
+    char buf[3200];
+    int nread, final;
+
+    VoskModel *model = vosk_model_new("model");
+    VoskSpkModel *spk_model = vosk_spk_model_new("spk-model");
+    VoskRecognizer *recognizer = vosk_recognizer_new_spk(model, 16000.0, spk_model);
+
+    wavin = fopen("test.wav", "rb");
+    fseek(wavin, 44, SEEK_SET);
+    while (!feof(wavin)) {
+         nread = fread(buf, 1, sizeof(buf), wavin);
+         final = vosk_recognizer_accept_waveform(recognizer, buf, nread);
+         if (final) {
+             printf("%s\n", vosk_recognizer_result(recognizer));
+         } else {
+             printf("%s\n", vosk_recognizer_partial_result(recognizer));
+         }
+    }
+    printf("%s\n", vosk_recognizer_final_result(recognizer));
+
+    vosk_recognizer_free(recognizer);
+    vosk_spk_model_free(spk_model);
+    vosk_model_free(model);
+    return 0;
+}

csharp/Makefile

@@ -1,55 +0,0 @@
-KALDI_ROOT ?= $(HOME)/kaldi
-CFLAGS :=  -std=c++11 -g -O2 -DPIC -fPIC -Wno-unused-function
-CPPFLAGS := -I$(KALDI_ROOT)/src -I$(KALDI_ROOT)/tools/openfst/include -I../src -DFST_NO_DYNAMIC_LINKING
-
-KALDI_LIBS = \
-             ${KALDI_ROOT}/src/online2/kaldi-online2.a \
-             ${KALDI_ROOT}/src/decoder/kaldi-decoder.a \
-             ${KALDI_ROOT}/src/ivector/kaldi-ivector.a \
-             ${KALDI_ROOT}/src/gmm/kaldi-gmm.a \
-             ${KALDI_ROOT}/src/nnet3/kaldi-nnet3.a \
-             ${KALDI_ROOT}/src/tree/kaldi-tree.a \
-             ${KALDI_ROOT}/src/feat/kaldi-feat.a \
-             ${KALDI_ROOT}/src/lat/kaldi-lat.a \
-             ${KALDI_ROOT}/src/lm/kaldi-lm.a \
-             ${KALDI_ROOT}/src/hmm/kaldi-hmm.a \
-             ${KALDI_ROOT}/src/transform/kaldi-transform.a \
-             ${KALDI_ROOT}/src/cudamatrix/kaldi-cudamatrix.a \
-             ${KALDI_ROOT}/src/matrix/kaldi-matrix.a \
-             ${KALDI_ROOT}/src/fstext/kaldi-fstext.a \
-             ${KALDI_ROOT}/src/util/kaldi-util.a \
-             ${KALDI_ROOT}/src/base/kaldi-base.a \
-             ${KALDI_ROOT}/tools/openfst/lib/libfst.a \
-             ${KALDI_ROOT}/tools/openfst/lib/libfstngram.a \
-             ${KALDI_ROOT}/tools/OpenBLAS/libopenblas.a \
-             -lgfortran -lstdc++
-
-all: test.exe
-
-test.exe: libkaldiwrap.so test.cs
-	mcs test.cs gen/*.cs
-
-VOSK_SOURCES = \
-	vosk_wrap.c \
-	../src/kaldi_recognizer.cc \
-	../src/kaldi_recognizer.h \
-	../src/model.cc \
-	../src/model.h \
-	../src/spk_model.cc \
-	../src/spk_model.h \
-	../src/vosk_api.cc \
-	../src/vosk_api.h
-
-libkaldiwrap.so: $(VOSK_SOURCES)
-	$(CXX) -fpermissive $(CFLAGS) $(CPPFLAGS) -shared -o $@ $(VOSK_SOURCES) $(KALDI_LIBS)
-
-vosk_wrap.c: ../src/vosk.i
-	mkdir -p gen
-	swig -csharp -DSWIG_CSHARP_NO_EXCEPTION_HELPER -dllimport "libkaldiwrap" \
-	    -namespace "Kaldi" -outdir gen -o vosk_wrap.c ../src/vosk.i
-
-run: test.exe
-	mono test.exe
-
-clean:
-	$(RM) *.so vosk_wrap.c *.o gen/*.cs test.exe

csharp/README.md

@@ -0,0 +1,12 @@
+This is a nuget-based wrapper for libvosk library
+
+See demo folder for example how to use the library. You can simply run
+"dotnet run" to run the demo. Make sure you unpacked the model and the
+test file.
+
+See the nuget folder for the sources of the wrapper. Run build.sh to
+build nuget package.
+
+Note we only support win64 and linux64 for now. No support for win32
+since it is a little bit painful to load the libraries depending on
+architecture. In theory we can add OSX some time or even Android.

csharp/demo/VoskDemo.cs

@@ -1,15 +1,57 @@
 using System;
 using System.IO;
-using Kaldi;
+using Vosk;
 
-public class Test
+public class VoskDemo
 {
-   public static void Main()
+   public static void DemoBytes(Model model)
    {
+        // Demo byte buffer
+        VoskRecognizer rec = new VoskRecognizer(model, 16000.0f);
+        rec.SetMaxAlternatives(0);
+        rec.SetWords(true);
+        using(Stream source = File.OpenRead("test.wav")) {
+            byte[] buffer = new byte[4096];
+            int bytesRead;
+            while((bytesRead = source.Read(buffer, 0, buffer.Length)) > 0) {
+                if (rec.AcceptWaveform(buffer, bytesRead)) {
+                    Console.WriteLine(rec.Result());
+                } else {
+                    Console.WriteLine(rec.PartialResult());
+                }
+            }
+        }
+        Console.WriteLine(rec.FinalResult());
+   }
 
-        Vosk.SetLogLevel(0);
-        Model model = new Model("model");
-        KaldiRecognizer rec = new KaldiRecognizer(model, 16000.0f);
+   public static void DemoFloats(Model model)
+   {
+        // Demo float array
+        VoskRecognizer rec = new VoskRecognizer(model, 16000.0f);
+        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);
+                }
+                if (rec.AcceptWaveform(fbuffer, fbuffer.Length)) {
+                    Console.WriteLine(rec.Result());
+                } else {
+                    Console.WriteLine(rec.PartialResult());
+                }
+            }
+        }
+        Console.WriteLine(rec.FinalResult());
+   }
+
+   public static void DemoSpeaker(Model model)
+   {
+        // Output speakers
+        SpkModel spkModel = new SpkModel("model-spk");
+        VoskRecognizer rec = new VoskRecognizer(model, 16000.0f);
+        rec.SetSpkModel(spkModel);
 
         using(Stream source = File.OpenRead("test.wav")) {
             byte[] buffer = new byte[4096];
@@ -23,25 +65,16 @@ public class Test
             }
         }
         Console.WriteLine(rec.FinalResult());
+   }
 
-        rec = new KaldiRecognizer(model, 16000.0f);
+   public static void Main()
+   {
+        // You can set to -1 to disable logging messages
+        Vosk.Vosk.SetLogLevel(0);
 
-        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);
-                }
-                if (rec.AcceptWaveform(fbuffer, fbuffer.Length)) {
-                    Console.WriteLine(rec.Result());
-                    GC.Collect();
-                } else {
-                    Console.WriteLine(rec.PartialResult());
-                }
-            }
-        }
-        Console.WriteLine(rec.FinalResult());
+        Model model = new Model("model");
+        DemoBytes(model);
+        DemoFloats(model);
+        DemoSpeaker(model);
    }
 }

csharp/demo/VoskDemo.csproj

@@ -0,0 +1,17 @@
+<Project Sdk="Microsoft.NET.Sdk">
+
+  <PropertyGroup>
+    <OutputType>Exe</OutputType>
+    <TargetFramework>net5.0</TargetFramework>
+    <RootNamespace>VoskDemo</RootNamespace>
+  </PropertyGroup>
+
+  <PropertyGroup>
+    <RestoreSources>$(RestoreSources);../nuget</RestoreSources>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <PackageReference Include="Vosk" Version="0.3.31" />
+  </ItemGroup>
+
+</Project>

csharp/nuget/Vosk.nuspec

@@ -0,0 +1,30 @@
+<?xml version="1.0"?>
+<package>
+  <metadata>
+    <id>Vosk</id>
+    <version>0.3.32</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.
+
+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++ 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.</description>
+    <releaseNotes>See for details https://github.com/alphacep/vosk-api/releases</releaseNotes>
+    <copyright>Copyright 2020 Alpha Cephei Inc</copyright>
+    <tags>speech recognition voice stt asr speech-to-text ai offline privacy</tags>
+    <dependencies>
+      <group targetFramework=".NETStandard2.0"/>
+    </dependencies>
+  </metadata>
+  <files>
+    <file src="**" exclude="src/*.cs;build.sh;**/.keep-me;*.nupkg" />
+  </files>
+</package>

csharp/nuget/build.sh

@@ -0,0 +1,2 @@
+mcs -out:lib/netstandard2.0/Vosk.dll -target:library src/*.cs
+nuget pack

csharp/nuget/build/Vosk.targets

@@ -0,0 +1,11 @@
+<Project xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <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))'" />
+    <None Include="@(NativeLibs)">
+      <Link>%(FileName)%(Extension)</Link>
+      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
+    </None>
+  </ItemGroup>
+</Project>

csharp/nuget/src/Model.cs

@@ -0,0 +1,41 @@
+namespace Vosk {
+
+public class Model : global::System.IDisposable {
+  private global::System.Runtime.InteropServices.HandleRef handle;
+
+  internal Model(global::System.IntPtr cPtr) {
+    handle = new global::System.Runtime.InteropServices.HandleRef(this, cPtr);
+  }
+
+  internal static global::System.Runtime.InteropServices.HandleRef getCPtr(Model obj) {
+    return (obj == null) ? new global::System.Runtime.InteropServices.HandleRef(null, global::System.IntPtr.Zero) : obj.handle;
+  }
+
+  ~Model() {
+    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_Model(handle);
+        handle = new global::System.Runtime.InteropServices.HandleRef(null, global::System.IntPtr.Zero);
+      }
+    }
+  }
+
+  public Model(string model_path) : this(VoskPINVOKE.new_Model(model_path)) {
+  }
+
+  public int vosk_model_find_word(string word) {
+    return VoskPINVOKE.Model_vosk_model_find_word(handle, word);
+  }
+
+}
+
+}

csharp/nuget/src/SpkModel.cs

@@ -0,0 +1,37 @@
+namespace Vosk {
+
+public class SpkModel : global::System.IDisposable {
+  private global::System.Runtime.InteropServices.HandleRef handle;
+
+  internal SpkModel(global::System.IntPtr cPtr) {
+    handle = new global::System.Runtime.InteropServices.HandleRef(this, cPtr);
+  }
+
+  internal static global::System.Runtime.InteropServices.HandleRef getCPtr(SpkModel obj) {
+    return (obj == null) ? new global::System.Runtime.InteropServices.HandleRef(null, global::System.IntPtr.Zero) : obj.handle;
+  }
+
+  ~SpkModel() {
+    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_SpkModel(handle);
+        handle = new global::System.Runtime.InteropServices.HandleRef(null, global::System.IntPtr.Zero);
+      }
+    }
+  }
+
+  public SpkModel(string model_path) : this(VoskPINVOKE.new_SpkModel(model_path)) {
+  }
+
+}
+
+}

csharp/nuget/src/Vosk.cs

@@ -0,0 +1,17 @@
+namespace Vosk {
+
+public class Vosk {
+  public static void SetLogLevel(int level) {
+    VoskPINVOKE.SetLogLevel(level);
+  }
+
+  public static void GpuInit() {
+    VoskPINVOKE.GpuInit();
+  }
+
+  public static void GpuThreadInit() {
+    VoskPINVOKE.GpuThreadInit();
+  }
+}
+
+}

csharp/nuget/src/VoskPINVOKE.cs

@@ -0,0 +1,75 @@
+namespace Vosk {
+
+class VoskPINVOKE {
+
+  static VoskPINVOKE() {
+  }
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_model_new")]
+  public static extern global::System.IntPtr new_Model(string jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_model_free")]
+  public static extern void delete_Model(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_model_find_word")]
+  public static extern int Model_vosk_model_find_word(global::System.Runtime.InteropServices.HandleRef jarg1, string jarg2);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_spk_model_new")]
+  public static extern global::System.IntPtr new_SpkModel(string jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_spk_model_free")]
+  public static extern void delete_SpkModel(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_new")]
+  public static extern global::System.IntPtr new_VoskRecognizer(global::System.Runtime.InteropServices.HandleRef jarg1, float jarg2);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_new_spk")]
+  public static extern global::System.IntPtr new_VoskRecognizerSpk(global::System.Runtime.InteropServices.HandleRef jarg1, float jarg2, global::System.Runtime.InteropServices.HandleRef jarg3);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_new_grm")]
+  public static extern global::System.IntPtr new_VoskRecognizerGrm(global::System.Runtime.InteropServices.HandleRef jarg1, float jarg2, string jarg3);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_free")]
+  public static extern void delete_VoskRecognizer(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_set_max_alternatives")]
+  public static extern void VoskRecognizer_SetMaxAlternatives(global::System.Runtime.InteropServices.HandleRef jarg1, int jarg2);
+
+  [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_spk_model")]
+  public static extern void VoskRecognizer_SetSpkModel(global::System.Runtime.InteropServices.HandleRef jarg1, global::System.Runtime.InteropServices.HandleRef jarg2);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_accept_waveform")]
+  public static extern bool VoskRecognizer_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_recognizer_accept_waveform_s")]
+  public static extern bool VoskRecognizer_AcceptWaveformShort(global::System.Runtime.InteropServices.HandleRef jarg1, [global::System.Runtime.InteropServices.In, global::System.Runtime.InteropServices.MarshalAs(global::System.Runtime.InteropServices.UnmanagedType.LPArray)]short[] jarg2, int jarg3);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_accept_waveform_f")]
+  public static extern bool VoskRecognizer_AcceptWaveformFloat(global::System.Runtime.InteropServices.HandleRef jarg1, [global::System.Runtime.InteropServices.In, global::System.Runtime.InteropServices.MarshalAs(global::System.Runtime.InteropServices.UnmanagedType.LPArray)]float[] jarg2, int jarg3);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_result")]
+  public static extern global::System.IntPtr VoskRecognizer_Result(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_partial_result")]
+  public static extern global::System.IntPtr VoskRecognizer_PartialResult(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_recognizer_final_result")]
+  public static extern global::System.IntPtr VoskRecognizer_FinalResult(global::System.Runtime.InteropServices.HandleRef jarg1);
+
+  [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_set_log_level")]
+  public static extern void SetLogLevel(int jarg1);
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_gpu_init")]
+  public static extern void GpuInit();
+
+  [global::System.Runtime.InteropServices.DllImport("libvosk", EntryPoint="vosk_gpu_thread_init")]
+  public static extern void GpuThreadInit();
+}
+
+}

csharp/nuget/src/VoskRecognizer.cs

@@ -0,0 +1,92 @@
+namespace Vosk {
+
+public class VoskRecognizer : System.IDisposable {
+  private System.Runtime.InteropServices.HandleRef handle;
+
+  internal VoskRecognizer(System.IntPtr cPtr) {
+    handle = new System.Runtime.InteropServices.HandleRef(this, cPtr);
+  }
+
+  internal static System.Runtime.InteropServices.HandleRef getCPtr(VoskRecognizer obj) {
+    return (obj == null) ? new System.Runtime.InteropServices.HandleRef(null, System.IntPtr.Zero) : obj.handle;
+  }
+
+  ~VoskRecognizer() {
+    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_VoskRecognizer(handle);
+        handle = new System.Runtime.InteropServices.HandleRef(null, System.IntPtr.Zero);
+      }
+    }
+  }
+
+  public VoskRecognizer(Model model, float sample_rate) : this(VoskPINVOKE.new_VoskRecognizer(Model.getCPtr(model), sample_rate)) {
+  }
+
+  public VoskRecognizer(Model model, float sample_rate, SpkModel spk_model) : this(VoskPINVOKE.new_VoskRecognizerSpk(Model.getCPtr(model), sample_rate, SpkModel.getCPtr(spk_model))) {
+  }
+
+  public VoskRecognizer(Model model, float sample_rate, string grammar) : this(VoskPINVOKE.new_VoskRecognizerGrm(Model.getCPtr(model), sample_rate, grammar)) {
+  }
+
+  public void SetMaxAlternatives(int max_alternatives) {
+    VoskPINVOKE.VoskRecognizer_SetMaxAlternatives(handle, max_alternatives);
+  }
+
+  public void SetWords(bool words) {
+    VoskPINVOKE.VoskRecognizer_SetWords(handle, words ? 1 : 0);
+  }
+
+  public void SetSpkModel(SpkModel spk_model) {
+    VoskPINVOKE.VoskRecognizer_SetSpkModel(handle, SpkModel.getCPtr(spk_model));
+  }
+
+  public bool AcceptWaveform(byte[] data, int len) {
+    return VoskPINVOKE.VoskRecognizer_AcceptWaveform(handle, data, len);
+  }
+
+  public bool AcceptWaveform(short[] sdata, int len) {
+    return VoskPINVOKE.VoskRecognizer_AcceptWaveformShort(handle, sdata, len);
+  }
+
+  public bool AcceptWaveform(float[] fdata, int len) {
+    return VoskPINVOKE.VoskRecognizer_AcceptWaveformFloat(handle, fdata, 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 Result() {
+    return PtrToStringUTF8(VoskPINVOKE.VoskRecognizer_Result(handle));
+  }
+
+  public string PartialResult() {
+    return PtrToStringUTF8(VoskPINVOKE.VoskRecognizer_PartialResult(handle));
+  }
+
+  public string FinalResult() {
+    return PtrToStringUTF8(VoskPINVOKE.VoskRecognizer_FinalResult(handle));
+  }
+
+  public void Reset() {
+    VoskPINVOKE.VoskRecognizer_Reset(handle);
+  }
+
+}
+
+}

doc/accuracy.md

@@ -1 +0,0 @@
-See https://alphacephei.com/vosk/accuracy.html

doc/adaptation.md

@@ -1 +0,0 @@
-See https://alphacephei.com/vosk/adaptation.html

doc/models.md

@@ -1 +0,0 @@
-See https://alphacephei.com/vosk/models.html

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/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/doc.go

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

go/example/test_simple.go

@@ -0,0 +1,56 @@
+package main
+
+import (
+	"bufio"
+	"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()
+
+	model, err := vosk.NewModel("model")
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	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()
+
+	reader := bufio.NewReader(file)
+	buf := make([]byte, 4096)
+
+	for {
+		_, err := reader.Read(buf)
+		if err != nil {
+			if err != io.EOF {
+				log.Fatal(err)
+			}
+
+			break
+		}
+
+		if rec.AcceptWaveform(buf) != 0 {
+			fmt.Println(string(rec.Result()))
+		}
+	}
+
+	fmt.Println(string(rec.FinalResult()))
+}

go/go.mod

@@ -0,0 +1,7 @@
+module github.com/alphacep/vosk-api/go
+
+go 1.16
+
+replace (
+    github.com/alphacep/vosk-api/go => ./
+)

go/vosk.go

@@ -0,0 +1,152 @@
+package vosk
+
+// #cgo CPPFLAGS: -I ${SRCDIR}/../src
+// #cgo LDFLAGS: -L ${SRCDIR}/../src -lvosk -ldl -lpthread
+// #include <stdlib.h>
+// #include <vosk_api.h>
+import "C"
+
+// VoskModel contains a reference to the C VoskModel
+type VoskModel struct {
+	model *C.struct_VoskModel
+}
+
+// NewModel creates a new VoskModel instance
+func NewModel(modelPath string) (*VoskModel, error) {
+	internal := C.vosk_model_new(C.CString(modelPath))
+	model := &VoskModel{model: internal}
+	return model, nil
+}
+
+func (m *VoskModel) Free() {
+	C.vosk_model_free(m.model)
+}
+
+func freeModel(model *VoskModel) {
+	C.vosk_model_free(model.model)
+}
+
+// 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 []byte) int {
+	cbuf := C.CBytes(word)
+	defer C.free(cbuf)
+	i := C.vosk_model_find_word(m.model, (*C.char)(cbuf))
+	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) {
+	internal := C.vosk_spk_model_new(C.CString(spkModelPath))
+	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, grammer []byte) (*VoskRecognizer, error) {
+	cbuf := C.CBytes(grammer)
+	defer C.free(cbuf)
+	internal := C.vosk_recognizer_new_grm(model.model, C.float(sampleRate), (*C.char)(cbuf))
+	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)
+}
+
+// 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))
+}
+
+// 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() []byte {
+	return []byte(C.GoString(C.vosk_recognizer_result(r.rec)))
+}
+
+// PartialResult returns a partial speech recognition result.
+func (r *VoskRecognizer) PartialResult() []byte {
+	return []byte(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() []byte {
+	return []byte(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

@@ -16,56 +16,12 @@
 		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 */; };
-		92375266240C6EFE00DD6076 /* disambig_tid.int in CopyFiles */ = {isa = PBXBuildFile; fileRef = 92375248240C6E3D00DD6076 /* disambig_tid.int */; };
-		92375267240C6EFE00DD6076 /* final.mdl in CopyFiles */ = {isa = PBXBuildFile; fileRef = 92375249240C6E3D00DD6076 /* final.mdl */; };
-		92375268240C6EFE00DD6076 /* Gr.fst in CopyFiles */ = {isa = PBXBuildFile; fileRef = 9237524A240C6E3D00DD6076 /* Gr.fst */; };
-		92375269240C6EFE00DD6076 /* HCLr.fst in CopyFiles */ = {isa = PBXBuildFile; fileRef = 9237524B240C6E3D00DD6076 /* HCLr.fst */; };
-		9237526A240C6EFE00DD6076 /* mfcc.conf in CopyFiles */ = {isa = PBXBuildFile; fileRef = 92375253240C6E3D00DD6076 /* mfcc.conf */; };
-		9237526B240C6EFE00DD6076 /* word_boundary.int in CopyFiles */ = {isa = PBXBuildFile; fileRef = 92375254240C6E3D00DD6076 /* word_boundary.int */; };
-		9237526C240C6EFE00DD6076 /* words.txt in CopyFiles */ = {isa = PBXBuildFile; fileRef = 92375255240C6E3D00DD6076 /* words.txt */; };
-		9237526E240C6F1500DD6076 /* final.dubm in CopyFiles */ = {isa = PBXBuildFile; fileRef = 9237524D240C6E3D00DD6076 /* final.dubm */; };
-		9237526F240C6F1500DD6076 /* final.ie in CopyFiles */ = {isa = PBXBuildFile; fileRef = 9237524E240C6E3D00DD6076 /* final.ie */; };
-		92375270240C6F1500DD6076 /* final.mat in CopyFiles */ = {isa = PBXBuildFile; fileRef = 9237524F240C6E3D00DD6076 /* final.mat */; };
-		92375271240C6F1500DD6076 /* global_cmvn.stats in CopyFiles */ = {isa = PBXBuildFile; fileRef = 92375250240C6E3D00DD6076 /* global_cmvn.stats */; };
-		92375272240C6F1500DD6076 /* online_cmvn.conf in CopyFiles */ = {isa = PBXBuildFile; fileRef = 92375251240C6E3D00DD6076 /* online_cmvn.conf */; };
-		92375273240C6F1500DD6076 /* splice.conf in CopyFiles */ = {isa = PBXBuildFile; fileRef = 92375252240C6E3D00DD6076 /* splice.conf */; };
 		92375274240C6F1E00DD6076 /* 10001-90210-01803.wav in Resources */ = {isa = PBXBuildFile; fileRef = 92375256240C6E3D00DD6076 /* 10001-90210-01803.wav */; };
+		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 */; };
 /* End PBXBuildFile section */
 
-/* Begin PBXCopyFilesBuildPhase section */
-		92375265240C6ECF00DD6076 /* CopyFiles */ = {
-			isa = PBXCopyFilesBuildPhase;
-			buildActionMask = 2147483647;
-			dstPath = "model-en";
-			dstSubfolderSpec = 7;
-			files = (
-				92375266240C6EFE00DD6076 /* disambig_tid.int in CopyFiles */,
-				92375267240C6EFE00DD6076 /* final.mdl in CopyFiles */,
-				92375268240C6EFE00DD6076 /* Gr.fst in CopyFiles */,
-				92375269240C6EFE00DD6076 /* HCLr.fst in CopyFiles */,
-				9237526A240C6EFE00DD6076 /* mfcc.conf in CopyFiles */,
-				9237526B240C6EFE00DD6076 /* word_boundary.int in CopyFiles */,
-				9237526C240C6EFE00DD6076 /* words.txt in CopyFiles */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-		9237526D240C6F0400DD6076 /* CopyFiles */ = {
-			isa = PBXCopyFilesBuildPhase;
-			buildActionMask = 2147483647;
-			dstPath = "model-en/ivector";
-			dstSubfolderSpec = 7;
-			files = (
-				9237526E240C6F1500DD6076 /* final.dubm in CopyFiles */,
-				9237526F240C6F1500DD6076 /* final.ie in CopyFiles */,
-				92375270240C6F1500DD6076 /* final.mat in CopyFiles */,
-				92375271240C6F1500DD6076 /* global_cmvn.stats in CopyFiles */,
-				92375272240C6F1500DD6076 /* online_cmvn.conf in CopyFiles */,
-				92375273240C6F1500DD6076 /* splice.conf in CopyFiles */,
-			);
-			runOnlyForDeploymentPostprocessing = 0;
-		};
-/* End PBXCopyFilesBuildPhase section */
-
 /* Begin PBXFileReference section */
 		9237521E240C550B00DD6076 /* VoskApiTest.app */ = {isa = PBXFileReference; explicitFileType = wrapper.application; includeInIndex = 0; path = VoskApiTest.app; sourceTree = BUILT_PRODUCTS_DIR; };
 		92375221240C550B00DD6076 /* AppDelegate.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = AppDelegate.swift; sourceTree = "<group>"; };
@@ -78,22 +34,12 @@
 		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; };
-		92375248240C6E3D00DD6076 /* disambig_tid.int */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = disambig_tid.int; sourceTree = "<group>"; };
-		92375249240C6E3D00DD6076 /* final.mdl */ = {isa = PBXFileReference; lastKnownFileType = file; path = final.mdl; sourceTree = "<group>"; };
-		9237524A240C6E3D00DD6076 /* Gr.fst */ = {isa = PBXFileReference; lastKnownFileType = file; path = Gr.fst; sourceTree = "<group>"; };
-		9237524B240C6E3D00DD6076 /* HCLr.fst */ = {isa = PBXFileReference; lastKnownFileType = file; path = HCLr.fst; sourceTree = "<group>"; };
-		9237524D240C6E3D00DD6076 /* final.dubm */ = {isa = PBXFileReference; lastKnownFileType = file; path = final.dubm; sourceTree = "<group>"; };
-		9237524E240C6E3D00DD6076 /* final.ie */ = {isa = PBXFileReference; lastKnownFileType = file; path = final.ie; sourceTree = "<group>"; };
-		9237524F240C6E3D00DD6076 /* final.mat */ = {isa = PBXFileReference; lastKnownFileType = file; path = final.mat; sourceTree = "<group>"; };
-		92375250240C6E3D00DD6076 /* global_cmvn.stats */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = global_cmvn.stats; sourceTree = "<group>"; };
-		92375251240C6E3D00DD6076 /* online_cmvn.conf */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = online_cmvn.conf; sourceTree = "<group>"; };
-		92375252240C6E3D00DD6076 /* splice.conf */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = splice.conf; sourceTree = "<group>"; };
-		92375253240C6E3D00DD6076 /* mfcc.conf */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = mfcc.conf; sourceTree = "<group>"; };
-		92375254240C6E3D00DD6076 /* word_boundary.int */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = word_boundary.int; sourceTree = "<group>"; };
-		92375255240C6E3D00DD6076 /* words.txt */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = words.txt; sourceTree = "<group>"; };
 		92375256240C6E3D00DD6076 /* 10001-90210-01803.wav */ = {isa = PBXFileReference; lastKnownFileType = audio.wav; path = "10001-90210-01803.wav"; sourceTree = "<group>"; };
+		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>"; };
+		92D6B8D225BDFEAC007FF08D /* VoskModel.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = VoskModel.swift; sourceTree = "<group>"; };
+		92D86BD4253F823F0040D53F /* vosk-model-spk-0.4 */ = {isa = PBXFileReference; lastKnownFileType = folder; path = "vosk-model-spk-0.4"; sourceTree = "<group>"; };
 /* End PBXFileReference section */
 
 /* Begin PBXFrameworksBuildPhase section */
@@ -139,6 +85,7 @@
 				9237522A240C550B00DD6076 /* LaunchScreen.storyboard */,
 				9237522D240C550B00DD6076 /* Info.plist */,
 				92375233240C558900DD6076 /* Vosk.swift */,
+				92D6B8D225BDFEAC007FF08D /* VoskModel.swift */,
 			);
 			path = VoskApiTest;
 			sourceTree = "<group>";
@@ -146,7 +93,8 @@
 		92375239240C642000DD6076 /* Vosk */ = {
 			isa = PBXGroup;
 			children = (
-				92375247240C6E3D00DD6076 /* model-android */,
+				928CC50C25BE124400490481 /* vosk-model-small-en-us-0.15 */,
+				92D86BD4253F823F0040D53F /* vosk-model-spk-0.4 */,
 				92375256240C6E3D00DD6076 /* 10001-90210-01803.wav */,
 				92AA22AD244CDD1200DA464B /* vosk_api.h */,
 				9237523A240C642000DD6076 /* libkaldiwrap.a */,
@@ -164,34 +112,6 @@
 			name = Frameworks;
 			sourceTree = "<group>";
 		};
-		92375247240C6E3D00DD6076 /* model-android */ = {
-			isa = PBXGroup;
-			children = (
-				92375248240C6E3D00DD6076 /* disambig_tid.int */,
-				92375249240C6E3D00DD6076 /* final.mdl */,
-				9237524A240C6E3D00DD6076 /* Gr.fst */,
-				9237524B240C6E3D00DD6076 /* HCLr.fst */,
-				9237524C240C6E3D00DD6076 /* ivector */,
-				92375253240C6E3D00DD6076 /* mfcc.conf */,
-				92375254240C6E3D00DD6076 /* word_boundary.int */,
-				92375255240C6E3D00DD6076 /* words.txt */,
-			);
-			path = "model-android";
-			sourceTree = "<group>";
-		};
-		9237524C240C6E3D00DD6076 /* ivector */ = {
-			isa = PBXGroup;
-			children = (
-				9237524D240C6E3D00DD6076 /* final.dubm */,
-				9237524E240C6E3D00DD6076 /* final.ie */,
-				9237524F240C6E3D00DD6076 /* final.mat */,
-				92375250240C6E3D00DD6076 /* global_cmvn.stats */,
-				92375251240C6E3D00DD6076 /* online_cmvn.conf */,
-				92375252240C6E3D00DD6076 /* splice.conf */,
-			);
-			path = ivector;
-			sourceTree = "<group>";
-		};
 /* End PBXGroup section */
 
 /* Begin PBXNativeTarget section */
@@ -202,8 +122,6 @@
 				9237521A240C550B00DD6076 /* Sources */,
 				9237521B240C550B00DD6076 /* Frameworks */,
 				9237521C240C550B00DD6076 /* Resources */,
-				92375265240C6ECF00DD6076 /* CopyFiles */,
-				9237526D240C6F0400DD6076 /* CopyFiles */,
 			);
 			buildRules = (
 			);
@@ -254,9 +172,11 @@
 			isa = PBXResourcesBuildPhase;
 			buildActionMask = 2147483647;
 			files = (
+				92BACED125BE125A00B5CC93 /* vosk-model-small-en-us-0.15 in Resources */,
 				92375274240C6F1E00DD6076 /* 10001-90210-01803.wav in Resources */,
 				9237522C240C550B00DD6076 /* LaunchScreen.storyboard in Resources */,
 				92375229240C550B00DD6076 /* Assets.xcassets in Resources */,
+				92D86BD6253F823F0040D53F /* vosk-model-spk-0.4 in Resources */,
 				92375227240C550B00DD6076 /* Main.storyboard in Resources */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
@@ -270,6 +190,7 @@
 			files = (
 				92375224240C550B00DD6076 /* ViewController.swift in Sources */,
 				92375222240C550B00DD6076 /* AppDelegate.swift in Sources */,
+				92D6B8D325BDFEAC007FF08D /* VoskModel.swift in Sources */,
 				92375234240C558900DD6076 /* Vosk.swift in Sources */,
 			);
 			runOnlyForDeploymentPostprocessing = 0;
@@ -412,7 +333,7 @@
 			buildSettings = {
 				ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
 				CLANG_ENABLE_MODULES = YES;
-				ENABLE_BITCODE = NO;
+				ENABLE_BITCODE = YES;
 				INFOPLIST_FILE = VoskApiTest/Info.plist;
 				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
 				LIBRARY_SEARCH_PATHS = (
@@ -434,7 +355,7 @@
 			buildSettings = {
 				ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
 				CLANG_ENABLE_MODULES = YES;
-				ENABLE_BITCODE = NO;
+				ENABLE_BITCODE = YES;
 				INFOPLIST_FILE = VoskApiTest/Info.plist;
 				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
 				LIBRARY_SEARCH_PATHS = (

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

@@ -2,6 +2,9 @@
 <Workspace
    version = "1.0">
    <FileRef
-      location = "self:VoskApiTest.xcodeproj">
+      location = "group:/Users/shmyrev/Documents/IOS/VoskApiTest/VoskApiTest/Vosk/vosk-model-small-en-us-0.15">
+   </FileRef>
+   <FileRef
+      location = "self:">
    </FileRef>
 </Workspace>

ios/VoskApiTest/Base.lproj/Main.storyboard

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="UTF-8"?>
-<document type="com.apple.InterfaceBuilder3.CocoaTouch.Storyboard.XIB" version="3.0" toolsVersion="13771" targetRuntime="iOS.CocoaTouch" propertyAccessControl="none" useAutolayout="YES" useTraitCollections="YES" colorMatched="YES" initialViewController="BYZ-38-t0r">
-    <device id="retina4_7" orientation="portrait">
+<document type="com.apple.InterfaceBuilder3.CocoaTouch.Storyboard.XIB" version="3.0" toolsVersion="13771" targetRuntime="iOS.CocoaTouch" propertyAccessControl="none" useAutolayout="YES" useTraitCollections="YES" colorMatched="YES" initialViewController="bdW-KL-Y8Z">
+    <device id="retina5_5" orientation="portrait">
         <adaptation id="fullscreen"/>
     </device>
     <dependencies>
@@ -10,23 +10,52 @@
     </dependencies>
     <scenes>
         <!--View Controller-->
-        <scene sceneID="tne-QT-ifu">
+        <scene sceneID="nEc-89-Iqu">
             <objects>
-                <viewController id="BYZ-38-t0r" customClass="ViewController" customModule="VoskApiTest" customModuleProvider="target" sceneMemberID="viewController">
-                    <textView key="view" clipsSubviews="YES" multipleTouchEnabled="YES" contentMode="scaleToFill" editable="NO" textAlignment="natural" id="CtX-mx-X98">
-                        <rect key="frame" x="0.0" y="0.0" width="375" height="667"/>
-                        <autoresizingMask key="autoresizingMask" flexibleMaxX="YES" flexibleMaxY="YES"/>
+                <viewController id="bdW-KL-Y8Z" customClass="ViewController" customModule="VoskApiTest" customModuleProvider="target" sceneMemberID="viewController">
+                    <layoutGuides>
+                        <viewControllerLayoutGuide type="top" id="Hyr-Dz-4mU"/>
+                        <viewControllerLayoutGuide type="bottom" id="w4A-5X-uBu"/>
+                    </layoutGuides>
+                    <view key="view" contentMode="scaleToFill" id="m5v-US-bvR">
+                        <rect key="frame" x="0.0" y="0.0" width="414" height="736"/>
+                        <autoresizingMask key="autoresizingMask" widthSizable="YES" heightSizable="YES"/>
+                        <subviews>
+                            <button opaque="NO" contentMode="scaleToFill" fixedFrame="YES" contentHorizontalAlignment="center" contentVerticalAlignment="center" buttonType="roundedRect" lineBreakMode="middleTruncation" translatesAutoresizingMaskIntoConstraints="NO" id="IaT-no-U3i" userLabel="Microphone">
+                                <rect key="frame" x="124" y="34" width="157" height="30"/>
+                                <autoresizingMask key="autoresizingMask" flexibleMaxX="YES" flexibleMaxY="YES"/>
+                                <state key="normal" title="Recognize Microphone"/>
+                                <connections>
+                                    <action selector="runRecognizeMicrohpone:" destination="bdW-KL-Y8Z" eventType="touchUpInside" id="hGB-lz-N2B"/>
+                                </connections>
+                            </button>
+                            <button opaque="NO" contentMode="scaleToFill" fixedFrame="YES" contentHorizontalAlignment="center" contentVerticalAlignment="center" buttonType="roundedRect" lineBreakMode="middleTruncation" translatesAutoresizingMaskIntoConstraints="NO" id="GC5-nT-FQR" userLabel="File">
+                                <rect key="frame" x="90" y="84" width="221" height="41"/>
+                                <autoresizingMask key="autoresizingMask" flexibleMaxX="YES" flexibleMaxY="YES"/>
+                                <state key="normal" title="Recognize File"/>
+                                <connections>
+                                    <action selector="runRecognizeFile:" destination="bdW-KL-Y8Z" eventType="touchUpInside" id="xp5-Yi-rnN"/>
+                                </connections>
+                            </button>
+                            <textView clipsSubviews="YES" multipleTouchEnabled="YES" contentMode="scaleToFill" fixedFrame="YES" text="Results here" textAlignment="natural" translatesAutoresizingMaskIntoConstraints="NO" id="w4X-cu-USq">
+                                <rect key="frame" x="11" y="112" width="383" height="569"/>
+                                <autoresizingMask key="autoresizingMask" flexibleMaxX="YES" flexibleMaxY="YES"/>
+                                <color key="backgroundColor" white="1" alpha="1" colorSpace="calibratedWhite"/>
+                                <fontDescription key="fontDescription" type="system" pointSize="14"/>
+                                <textInputTraits key="textInputTraits" autocapitalizationType="sentences"/>
+                            </textView>
+                        </subviews>
                         <color key="backgroundColor" white="1" alpha="1" colorSpace="calibratedWhite"/>
-                        <fontDescription key="fontDescription" type="system" pointSize="14"/>
-                        <textInputTraits key="textInputTraits" autocapitalizationType="sentences"/>
-                    </textView>
+                    </view>
                     <connections>
-                        <outlet property="mainText" destination="CtX-mx-X98" id="oJy-5J-NKp"/>
+                        <outlet property="mainText" destination="w4X-cu-USq" id="rZS-nz-Wql"/>
+                        <outlet property="recognizeFile" destination="GC5-nT-FQR" id="dRe-tc-IA0"/>
+                        <outlet property="recognizeMicrophone" destination="IaT-no-U3i" id="IuM-aa-pAP"/>
                     </connections>
                 </viewController>
-                <placeholder placeholderIdentifier="IBFirstResponder" id="dkx-z0-nzr" sceneMemberID="firstResponder"/>
+                <placeholder placeholderIdentifier="IBFirstResponder" id="nWA-4D-pA6" userLabel="First Responder" sceneMemberID="firstResponder"/>
             </objects>
-            <point key="canvasLocation" x="32.799999999999997" y="32.833583208395808"/>
+            <point key="canvasLocation" x="-17.39130434782609" y="-285.32608695652175"/>
         </scene>
     </scenes>
 </document>

ios/VoskApiTest/ViewController.swift

@@ -3,30 +3,113 @@
 //  VoskApiTest
 //
 //  Created by Niсkolay Shmyrev on 01.03.20.
-//  Copyright © 2020 Alpha Cephei. All rights reserved.
+//  Copyright © 2020-2021 Alpha Cephei. All rights reserved.
 //
 
 import UIKit
+import AVFoundation
+
+enum WorkMode {
+    case stopped
+    case microphone
+    case file
+}
 
 class ViewController: UIViewController {
-
-    @IBOutlet var mainText: UITextView!
-
-    override func viewDidLoad() {
-        super.viewDidLoad()
     
-        DispatchQueue.global(qos: .userInitiated).async {
-            DispatchQueue.main.async {
-                 self.mainText.text =  "Processing file..."
+    var mode: WorkMode!
+    
+    @IBOutlet weak var recognizeFile: UIButton!
+    @IBOutlet weak var mainText: UITextView!
+    @IBOutlet weak var recognizeMicrophone: UIButton!
+    
+    var audioEngine : AVAudioEngine!
+    var processingQueue: DispatchQueue!
+    var model : VoskModel!
+    
+    func setMode(mode: WorkMode) {
+        switch mode {
+        case .stopped:
+            self.recognizeFile.isEnabled = true
+            self.recognizeMicrophone.isEnabled = true
+            self.recognizeMicrophone.setTitle("Recognize Microphone",for: .normal)
+        case .microphone:
+            self.recognizeFile.isEnabled = false
+            self.recognizeMicrophone.isEnabled = true
+            self.recognizeMicrophone.setTitle("Stop Microphone",for: .normal)
+            self.mainText.text = ""
+        case .file:
+            self.recognizeFile.isEnabled = false
+            self.recognizeMicrophone.isEnabled = false
+            self.mainText.text =  "Processing file..."
+        }
+        self.mode = mode
+    }
+    
+    func startAudioEngine() {
+        do {
+            
+            // Create a new audio engine.
+            audioEngine = AVAudioEngine()
+            
+            let inputNode = audioEngine.inputNode
+            let formatInput = inputNode.inputFormat(forBus: 0)
+            let formatPcm = AVAudioFormat.init(commonFormat: AVAudioCommonFormat.pcmFormatInt16, sampleRate: formatInput.sampleRate, channels: 1, interleaved: true)
+            
+            let recognizer = Vosk(model: model, sampleRate: Float(formatInput.sampleRate))
+            
+            inputNode.installTap(onBus: 0,
+                                 bufferSize: UInt32(formatInput.sampleRate / 10),
+                                 format: formatPcm) { buffer, time in
+                                    self.processingQueue.async {
+                                        let res = recognizer.recognizeData(buffer: buffer)
+                                        DispatchQueue.main.async {
+                                            self.mainText.text = res + "\n" + self.mainText.text
+                                        }
+                                    }
             }
-            let vosk = Vosk()
-            let res = vosk.recognizeFile()
+            
+            // Start the stream of audio data.
+            audioEngine.prepare()
+            try audioEngine.start()
+        } catch {
+            print("Unable to start AVAudioEngine: \(error.localizedDescription)")
+        }
+    }
+    
+    func stopAudioEngine() {
+        audioEngine.stop()
+    }
+    
+    @IBAction func runRecognizeMicrohpone(_ sender: Any) {
+        if (mode == .stopped) {
+            setMode(mode: .microphone)
+            startAudioEngine()
+        } else {
+            stopAudioEngine()
+            setMode(mode: .stopped)
+        }
+    }
+    
+    @IBAction func runRecognizeFile(_ sender: Any) {
+        setMode(mode: .file)
+        processingQueue.async {
+            let recognizer = Vosk(model: self.model, sampleRate: 16000.0)
+            let res = recognizer.recognizeFile()
             DispatchQueue.main.async {
                 self.mainText.text = res
+                self.setMode(mode: .stopped)
             }
         }
     }
-
+    
+    override func viewDidLoad() {
+        super.viewDidLoad()
+        setMode(mode: .stopped)
+        processingQueue = DispatchQueue(label: "recognizerQueue")
+        model = VoskModel()
+    }
+    
     override func didReceiveMemoryWarning() {
         super.didReceiveMemoryWarning()
     }

ios/VoskApiTest/Vosk.swift

@@ -3,35 +3,52 @@
 //  VoskApiTest
 //
 //  Created by Niсkolay Shmyrev on 01.03.20.
-//  Copyright © 2020 Alpha Cephei. All rights reserved.
+//  Copyright © 2020-2021 Alpha Cephei. All rights reserved.
 //
 
 import Foundation
+import AVFoundation
 
 public final class Vosk {
     
+    var recognizer : OpaquePointer!
+
+    init(model: VoskModel, sampleRate: Float) {
+        recognizer = vosk_recognizer_new_spk(model.model, model.spkModel, sampleRate)
+    }
+    
+    deinit {
+        vosk_recognizer_free(recognizer);
+    }
+    
     func recognizeFile() -> String {
         var sres = ""
+        
         if let resourcePath = Bundle.main.resourcePath {
-
-            let modelPath = resourcePath + "/model-en"
-
-            let model = vosk_model_new(modelPath);
-            let recognizer = vosk_recognizer_new(model, 16000.0)
-
+            
             let audioFile = URL(fileURLWithPath: resourcePath + "/10001-90210-01803.wav")
+    
             if let data = try? Data(contentsOf: audioFile) {
-                let _ = data.withUnsafeBytes {
-                    vosk_recognizer_accept_waveform(recognizer, $0, Int32(data.count))
-                }
-                let res = vosk_recognizer_final_result(recognizer);
-                sres = String(validatingUTF8: res!)!;
-                print(sres);
+                    let _ = data.withUnsafeBytes {
+                        vosk_recognizer_accept_waveform(recognizer, $0, Int32(data.count))
+                    }
+                    let res = vosk_recognizer_final_result(recognizer);
+                    sres = String(validatingUTF8: res!)!;
+                    print(sres);
             }
-
-            vosk_recognizer_free(recognizer)
-            vosk_model_free(model)
         }
+        
         return sres
     }
+    
+    
+    func recognizeData(buffer : AVAudioPCMBuffer) -> String {
+        let dataLen = Int(buffer.frameLength * 2)
+        let channels = UnsafeBufferPointer(start: buffer.int16ChannelData, count: 1)
+        let endOfSpeech = channels[0].withMemoryRebound(to: Int8.self, capacity: dataLen) {
+            vosk_recognizer_accept_waveform(recognizer, $0, Int32(dataLen))
+        }
+        let res = endOfSpeech == 1 ?vosk_recognizer_result(recognizer) :vosk_recognizer_partial_result(recognizer)
+        return String(validatingUTF8: res!)!;
+    }
 }

ios/VoskApiTest/Vosk/vosk_api.h

@@ -12,37 +12,200 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+/* This header contains the C API for Vosk speech recognition system */
 
-#ifndef _VOSK_API_H_
-#define _VOSK_API_H_
+#ifndef VOSK_API_H
+#define VOSK_API_H
 
 #ifdef __cplusplus
 extern "C" {
 #endif
 
+/** Model stores all the data required for recognition
+ *  it contains static data and can be shared across processing
+ *  threads. */
 typedef struct VoskModel VoskModel;
+
+
+/** Speaker model is the same as model but contains the data
+ *  for speaker identification. */
 typedef struct VoskSpkModel VoskSpkModel;
+
+
+/** 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 */
 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 */
 VoskModel *vosk_model_new(const char *model_path);
+
+
+/** 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. */
 void vosk_model_free(VoskModel *model);
 
+
+/** 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 */
 VoskSpkModel *vosk_spk_model_new(const char *model_path);
+
+
+/** 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. */
 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 */
 VoskRecognizer *vosk_recognizer_new(VoskModel *model, float sample_rate);
+
+
+/** 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 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);
+
+
+/** 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 sample_rate The sample rate of the audio you going to feed into the recognizer
+ *  @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 */
 VoskRecognizer *vosk_recognizer_new_grm(VoskModel *model, float sample_rate, const char *grammar);
+
+
+/** 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 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);
+
+
+/** 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>
+ * {
+ *   "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"
+ *     }],
+ *   "text" : "what zero zero zero one"
+ *  }
+ * </pre>
+ */
 const char *vosk_recognizer_result(VoskRecognizer *recognizer);
+
+
+/** 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>
+ */
 const char *vosk_recognizer_partial_result(VoskRecognizer *recognizer);
+
+
+/** 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.
+ */
 const char *vosk_recognizer_final_result(VoskRecognizer *recognizer);
+
+
+/** Releases recognizer object
+ *
+ *  Underlying model is also unreferenced and if needed released */
 void vosk_recognizer_free(VoskRecognizer *recognizer);
 
+/** Set log level for Kaldi messages
+ *
+ *  @param log_level the level
+ *     0 - default value to print info and error messages but no debug
+ *     less than 0 - don't print info messages
+ *     greather than 0 - more verbose mode
+ */
+void vosk_set_log_level(int log_level);
+
 #ifdef __cplusplus
 }
 #endif
 
-#endif /* _VOSK_API_H_ */
+#endif /* VOSK_API_H */

ios/VoskApiTest/VoskModel.swift

@@ -0,0 +1,36 @@
+//
+//  Vosk.swift
+//  VoskApiTest
+//
+//  Created by Niсkolay Shmyrev on 01.03.20.
+//  Copyright © 2020-2021 Alpha Cephei. All rights reserved.
+//
+
+import Foundation
+
+public final class VoskModel {
+    
+    var model : OpaquePointer!
+    var spkModel : OpaquePointer!
+    
+    init() {
+        
+        // Set to -1 to disable logs
+        vosk_set_log_level(0);
+        
+        if let resourcePath = Bundle.main.resourcePath {
+            let modelPath = resourcePath + "/vosk-model-small-en-us-0.15"
+            let spkModelPath = resourcePath + "/vosk-model-spk-0.4"
+            
+            model = vosk_model_new(modelPath)
+            spkModel = vosk_spk_model_new(spkModelPath)
+        }
+    }
+    
+    deinit {
+        vosk_model_free(model)
+        vosk_spk_model_free(spkModel)
+    }
+    
+}
+

java/Makefile

@@ -1,65 +0,0 @@
-KALDI_ROOT ?= $(HOME)/kaldi
-CFLAGS := -g -O2 -DPIC -fPIC -Wno-unused-function
-CPPFLAGS := -I$(JAVA_HOME)/include -I$(JAVA_HOME)/include/linux -I$(KALDI_ROOT)/src -I$(KALDI_ROOT)/tools/openfst/include -I../src
-
-KALDI_LIBS = \
-             ${KALDI_ROOT}/src/online2/kaldi-online2.a \
-             ${KALDI_ROOT}/src/decoder/kaldi-decoder.a \
-             ${KALDI_ROOT}/src/ivector/kaldi-ivector.a \
-             ${KALDI_ROOT}/src/gmm/kaldi-gmm.a \
-             ${KALDI_ROOT}/src/nnet3/kaldi-nnet3.a \
-             ${KALDI_ROOT}/src/tree/kaldi-tree.a \
-             ${KALDI_ROOT}/src/feat/kaldi-feat.a \
-             ${KALDI_ROOT}/src/lat/kaldi-lat.a \
-             ${KALDI_ROOT}/src/lm/kaldi-lm.a \
-             ${KALDI_ROOT}/src/hmm/kaldi-hmm.a \
-             ${KALDI_ROOT}/src/transform/kaldi-transform.a \
-             ${KALDI_ROOT}/src/cudamatrix/kaldi-cudamatrix.a \
-             ${KALDI_ROOT}/src/matrix/kaldi-matrix.a \
-             ${KALDI_ROOT}/src/fstext/kaldi-fstext.a \
-             ${KALDI_ROOT}/src/util/kaldi-util.a \
-             ${KALDI_ROOT}/src/base/kaldi-base.a \
-             ${KALDI_ROOT}/tools/openfst/lib/libfst.a \
-             ${KALDI_ROOT}/tools/openfst/lib/libfstngram.a \
-             ${KALDI_ROOT}/tools/OpenBLAS/libopenblas.a \
-             -lgfortran
-
-all: libvosk_jni.so
-
-VOSK_SOURCES = \
-	vosk_wrap.cc \
-	../src/kaldi_recognizer.cc \
-	../src/kaldi_recognizer.h \
-	../src/model.cc \
-	../src/model.h \
-	../src/spk_model.cc \
-	../src/spk_model.h \
-	../src/vosk_api.cc \
-	../src/vosk_api.h
-
-libvosk_jni.so: $(VOSK_SOURCES)
-	$(CXX) -shared -o $@ $(CPPFLAGS) $(CFLAGS) $(VOSK_SOURCES) $(KALDI_LIBS)
-
-vosk_wrap.cc: ../src/vosk.i
-	mkdir -p org/kaldi
-	swig -c++ -I../src \
-		-java -package org.kaldi \
-		-outdir org/kaldi -o $@ $<
-
-clean:
-	$(RM) *.so *_wrap.cc *_wrap.o test/*.class
-	$(RM) -r org model-en
-
-model:
-	wget https://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.zip
-	unzip vosk-model-small-en-us-0.3.zip && rm vosk-model-small-en-us-0.3.zip
-	mv vosk-model-small-en-us-0.3 model
-
-model-spk:
-	wget https://alphacephei.com/kaldi/models/vosk-model-spk-0.3.zip
-	unzip vosk-model-spk-0.3.zip && rm vosk-model-spk-0.3.zip
-	mv vosk-model-spk-0.3 model-spk
-
-run: model model-spk
-	javac test/*.java org/kaldi/*.java
-	java -Djava.library.path=. -cp . test.DecoderTest

java/README.md

@@ -1,19 +1,7 @@
-Java API sample
+Java bindings for Vosk API using jnr-ffi
 
-Doesn't work on Windows or Mac yet, help to prepare the packaged jars is welcome.
+See demo project for details, build it with Gradle.
 
-For now to try it:
+Download model and unpack as "model" folder in the demo project.
 
-On Linux you can do
-
-   1. Build recent kaldi
-   1. `git clone https://github.com/alphacep/vosk-api`
-   1. `cd vosk-api/java`
-   1. `export KALDI_ROOT=<KALDI_ROOT>`
-   1. `export JAVA_HOME=<JAVA_HOME>`
-   1. `make`
-   1. `make run`
-
-For details of the code you can check:
-
-https://github.com/alphacep/vosk-api/blob/master/java/test/DecoderTest.java
+Make sure you are using recent JDK and Gradle.

java/demo/build.gradle

@@ -0,0 +1,19 @@
+plugins {
+    id 'application'
+}
+
+application {
+    mainClass = 'org.vosk.demo.DecoderDemo'
+}
+
+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: '0.3.32+'
+}

java/demo/src/main/java/org/vosk/demo/DecoderDemo.java

@@ -0,0 +1,38 @@
+package org.vosk.demo;
+
+import java.io.FileInputStream;
+import java.io.BufferedInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+
+import javax.sound.sampled.AudioSystem;
+import javax.sound.sampled.UnsupportedAudioFileException;
+
+import org.vosk.LogLevel;
+import org.vosk.Recognizer;
+import org.vosk.LibVosk;
+import org.vosk.Model;
+
+public class DecoderDemo {
+
+    public static void main(String[] argv) throws IOException, UnsupportedAudioFileException {
+        LibVosk.setLogLevel(LogLevel.DEBUG);
+
+        try (Model model = new Model("model");
+                    InputStream ais = AudioSystem.getAudioInputStream(new BufferedInputStream(new FileInputStream("../../python/example/test.wav")));
+                    Recognizer recognizer = new Recognizer(model, 16000)) {
+
+            int nbytes;
+            byte[] b = new byte[4096];
+            while ((nbytes = ais.read(b)) >= 0) {
+                if (recognizer.acceptWaveForm(b, nbytes)) {
+                    System.out.println(recognizer.getResult());
+                } else {
+                    System.out.println(recognizer.getPartialResult());
+                }
+            }
+
+            System.out.println(recognizer.getFinalResult());
+        }
+    }
+}

java/lib/build.gradle

@@ -0,0 +1,58 @@
+plugins {
+    id 'java-library'
+    id 'maven-publish'
+}
+
+archivesBaseName = 'vosk'
+group = 'com.alphacephei'
+version = '0.3.32'
+
+repositories {
+    mavenCentral()
+}
+
+dependencies {
+    implementation group: 'net.java.dev.jna', name: 'jna', version: '5.7.0'
+    testImplementation 'junit:junit:4.13'
+}
+
+publishing {
+    publications {
+        mavenJava(MavenPublication) {
+            artifactId = 'vosk'
+            from components.java
+            pom {
+                name = 'Vosk'
+                description = 'Speech recognition library'
+                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 = '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/'
+                }
+            }
+        }
+    }
+    repositories {
+        maven {
+            url = "repo"
+        }
+    }
+}
+
+test {
+    dependsOn cleanTest
+    testLogging.showStandardStreams = true
+}

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

@@ -0,0 +1,86 @@
+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;
+import java.io.InputStream;
+import java.io.IOException;
+import java.nio.file.Files;
+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);
+    }
+
+    static {
+
+        if (Platform.isWindows()) {
+            // 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 tmpDir = tmpFile.getParentFile();
+                new File(tmpDir, tmpFile.getName() + ".x").createNewFile();
+
+                // Now unpack dependencies
+                unpackDll(tmpDir, "libwinpthread-1");
+                unpackDll(tmpDir, "libgcc_s_seh-1");
+                unpackDll(tmpDir, "libstdc++-6");
+
+            } catch (IOException e) {
+                // Nothing for now, it will fail on next step
+            } finally {
+                Native.register(LibVosk.class, "libvosk");
+            }
+        } else {
+            Native.register(LibVosk.class, "vosk");
+        }
+    }
+
+    public static native void vosk_set_log_level(int level);
+
+    public static native Pointer vosk_model_new(String path);
+
+    public static native void vosk_model_free(Pointer model);
+
+    public static native Pointer vosk_spk_model_new(String path);
+
+    public static native void vosk_spk_model_free(Pointer model);
+
+    public static native Pointer vosk_recognizer_new(Model model, float sample_rate);
+
+    public static native Pointer vosk_recognizer_new_spk(Pointer model, float sample_rate, Pointer spk_model);
+
+    public static native Pointer vosk_recognizer_new_grm(Pointer model, float sample_rate, String grammar);
+
+    public static native void vosk_recognizer_set_max_alternatives(Pointer recognizer, int max_alternatives);
+
+    public static native void vosk_recognizer_set_words(Pointer recognizer, boolean 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);
+
+    public static native boolean vosk_recognizer_accept_waveform_s(Pointer recognizer, short[] data, int len);
+
+    public static native boolean vosk_recognizer_accept_waveform_f(Pointer recognizer, float[] data, int len);
+
+    public static native String vosk_recognizer_result(Pointer recognizer);
+
+    public static native String vosk_recognizer_final_result(Pointer recognizer);
+
+    public static native String vosk_recognizer_partial_result(Pointer recognizer);
+
+    public static native void vosk_recognizer_reset(Pointer recognizer);
+
+    public static native void vosk_recognizer_free(Pointer recognizer);
+
+    public static void setLogLevel(LogLevel loglevel) {
+        vosk_set_log_level(loglevel.getValue());
+    }
+}

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

@@ -0,0 +1,17 @@
+package org.vosk;
+
+public enum LogLevel {
+    WARNINGS(-1),  // Print warning and errors
+    INFO(0),       // Print info, along with warning and error messages, but no debug
+    DEBUG(1);      // Print debug info
+
+    private final int value;
+
+    LogLevel(int value) {
+        this.value = value;
+    }
+
+    public int getValue() {
+        return this.value;
+    }
+}

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

@@ -0,0 +1,22 @@
+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) throws IOException {
+        super(LibVosk.vosk_model_new(path));
+
+        if (getPointer() == null) {
+            throw new IOException("Failed to create a model");
+        }
+    }
+
+    @Override
+    public void close() {
+        LibVosk.vosk_model_free(this.getPointer());
+    }
+}

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

@@ -0,0 +1,67 @@
+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) 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) {
+        super(LibVosk.vosk_recognizer_new_spk(model.getPointer(), sampleRate, spkModel.getPointer()));
+    }
+
+    public Recognizer(Model model, float sampleRate, String grammar) {
+        super(LibVosk.vosk_recognizer_new_grm(model.getPointer(), sampleRate, grammar));
+    }
+
+    public void setMaxAlternatives(int maxAlternatives) {
+        LibVosk.vosk_recognizer_set_max_alternatives(this.getPointer(), maxAlternatives);
+    }
+
+    public void setWords(boolean words) {
+        LibVosk.vosk_recognizer_set_words(this.getPointer(), words);
+    }
+
+    public void setSpeakerModel(SpeakerModel spkModel) {
+        LibVosk.vosk_recognizer_set_spk_model(this.getPointer(), spkModel.getPointer());
+    }
+
+    public boolean acceptWaveForm(byte[] data, int len) {
+        return LibVosk.vosk_recognizer_accept_waveform(this.getPointer(), data, len);
+    }
+
+    public boolean acceptWaveForm(short[] data, int len) {
+        return LibVosk.vosk_recognizer_accept_waveform_s(this.getPointer(), data, len);
+    }
+
+    public boolean acceptWaveForm(float[] data, int len) {
+        return LibVosk.vosk_recognizer_accept_waveform_f(this.getPointer(), data, len);
+    }
+
+    public String getResult() {
+        return LibVosk.vosk_recognizer_result(this.getPointer());
+    }
+
+    public String getPartialResult() {
+        return LibVosk.vosk_recognizer_partial_result(this.getPointer());
+    }
+
+    public String getFinalResult() {
+        return LibVosk.vosk_recognizer_final_result(this.getPointer());
+    }
+
+    public void reset() {
+        LibVosk.vosk_recognizer_reset(this.getPointer());
+    }
+
+    @Override
+    public void close() {
+        LibVosk.vosk_recognizer_free(this.getPointer());
+    }
+}

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

@@ -0,0 +1,22 @@
+package org.vosk;
+
+import com.sun.jna.PointerType;
+import java.io.IOException;
+
+public class SpeakerModel extends PointerType implements AutoCloseable {
+    public SpeakerModel() {
+    }
+
+    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
+    public void close() {
+        LibVosk.vosk_spk_model_free(this.getPointer());
+    }
+}

java/lib/src/main/resources/win32-x86-64/empty

@@ -0,0 +1 @@
+Mainly for work around JNA API

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

@@ -0,0 +1,102 @@
+package org.vosk.test;
+
+import java.io.FileInputStream;
+import java.io.BufferedInputStream;
+import java.io.IOException;
+import java.io.InputStream;
+import java.nio.ByteBuffer;
+import java.nio.ByteOrder;
+
+import org.junit.Test;
+import org.junit.Assert;
+
+import javax.sound.sampled.AudioSystem;
+import javax.sound.sampled.UnsupportedAudioFileException;
+
+import org.vosk.LogLevel;
+import org.vosk.Recognizer;
+import org.vosk.LibVosk;
+import org.vosk.Model;
+
+public class DecoderTest {
+
+    @Test
+    public void decoderTest() throws IOException, UnsupportedAudioFileException {
+        LibVosk.setLogLevel(LogLevel.DEBUG);
+
+        try (Model model = new Model("model");
+            InputStream ais = AudioSystem.getAudioInputStream(new BufferedInputStream(new FileInputStream("../../python/example/test.wav")));
+            Recognizer recognizer = new Recognizer(model, 16000)) {
+
+            recognizer.setMaxAlternatives(10);
+            recognizer.setWords(true);
+
+            int nbytes;
+            byte[] b = new byte[4096];
+            while ((nbytes = ais.read(b)) >= 0) {
+                if (recognizer.acceptWaveForm(b, nbytes)) {
+                    System.out.println(recognizer.getResult());
+                } else {
+                    System.out.println(recognizer.getPartialResult());
+                }
+            }
+
+            System.out.println(recognizer.getFinalResult());
+        }
+        Assert.assertTrue(true);
+    }
+
+    @Test
+    public void decoderTestShort() throws IOException, UnsupportedAudioFileException {
+        LibVosk.setLogLevel(LogLevel.DEBUG);
+
+        try (Model model = new Model("model");
+            InputStream ais = AudioSystem.getAudioInputStream(new BufferedInputStream(new FileInputStream("../../python/example/test.wav")));
+            Recognizer recognizer = new Recognizer(model, 16000)) {
+
+            int nbytes;
+            byte[] b = new byte[4096];
+            short[] s = new short[2048];
+            while ((nbytes = ais.read(b)) >= 0) {
+                ByteBuffer.wrap(b).order(ByteOrder.LITTLE_ENDIAN).asShortBuffer().get(s);
+                if (recognizer.acceptWaveForm(s, nbytes / 2)) {
+                    System.out.println(recognizer.getResult());
+                } else {
+                    System.out.println(recognizer.getPartialResult());
+                }
+            }
+
+            System.out.println(recognizer.getFinalResult());
+        }
+        Assert.assertTrue(true);
+    }
+
+    @Test
+    public void decoderTestGrammar() throws IOException, UnsupportedAudioFileException {
+        LibVosk.setLogLevel(LogLevel.DEBUG);
+
+        try (Model model = new Model("model");
+            InputStream ais = AudioSystem.getAudioInputStream(new BufferedInputStream(new FileInputStream("../../python/example/test.wav")));
+            Recognizer recognizer = new Recognizer(model, 16000, "[\"one two three four five six seven eight nine zero oh\"]")) {
+
+            int nbytes;
+            byte[] b = new byte[4096];
+            while ((nbytes = ais.read(b)) >= 0) {
+                if (recognizer.acceptWaveForm(b, nbytes)) {
+                    System.out.println(recognizer.getResult());
+                } else {
+                    System.out.println(recognizer.getPartialResult());
+                }
+            }
+
+            System.out.println(recognizer.getFinalResult());
+        }
+        Assert.assertTrue(true);
+    }
+
+
+    @Test(expected = IOException.class)
+    public void decoderTestException() throws IOException {
+        Model model = new Model("model_missing");
+    }
+}

java/test/DecoderTest.java

@@ -1,40 +0,0 @@
-package test;
-
-import java.io.File;
-import java.io.FileInputStream;
-import java.io.FileOutputStream;
-import java.io.DataOutputStream;
-import java.io.IOException;
-import java.net.URL;
-import java.nio.*;
-
-import org.kaldi.KaldiRecognizer;
-import org.kaldi.Model;
-import org.kaldi.SpkModel;
-import org.kaldi.Vosk;
-
-public class DecoderTest {
-    static {
-        System.loadLibrary("vosk_jni");
-    }
-
-    public static void main(String args[]) throws IOException {
-        Vosk.SetLogLevel(-10);
-
-        FileInputStream ais = new FileInputStream(new File("../python/example/test.wav"));
-        Model model = new Model("model");
-        SpkModel spkModel = new SpkModel("model-spk");
-        KaldiRecognizer rec = new KaldiRecognizer(model, spkModel, 16000.0f);
-
-        int nbytes;
-        byte[] b = new byte[4096];
-        while ((nbytes = ais.read(b)) >= 0) {
-            if (rec.AcceptWaveform(b)) {
-                System.out.println(rec.Result());
-            } else {
-                System.out.println(rec.PartialResult());
-            }
-        }
-        System.out.println(rec.FinalResult());
-    }
-}

nodejs/.npmignore

@@ -1,3 +1,3 @@
-build
-node_modules
-vosk_wrap.cc
+demo/model
+demo/model-spk
+demo/test.wav

nodejs/README.md

@@ -1,20 +1,33 @@
-Installation requires vosk-api checkout, it doesn't yet work with `npm
-install vosk`. We have to figure out how to properly distribute native
-modules for Vosk.
+This is an FFI-NAPI wrapper for the Vosk library.
 
-The build tested with node-0.10.15, node-0.12 is not yet supported by swig.
+## Usage
 
-Still, you need swig of newest version 4.0.1
+It mostly follows Vosk interface, some methods are not yet fully implemented.
 
-Build like this
+To use it you need to compile libvosk library, see Python module build
+instructions for details. You can find prebuilt library inside python
+wheel.
 
-```
-npm install --kaldi_root=/home/suser/kaldi
-```
+## About
 
-Then test with
+Vosk is an offline open source speech recognition toolkit. It enables
+speech recognition models for 17 languages and dialects - English, Indian
+English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish,
+Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi, Filipino.
 
-```
-cd example
-node test.js
-```
+Vosk models are small (50 Mb) but provide continuous large vocabulary
+transcription, zero-latency response with streaming API, reconfigurable
+vocabulary and speaker identification.
+
+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). See also our project on
+[Github](https://github.com/alphacep/vosk-api).

nodejs/binding.gyp

@@ -1,70 +0,0 @@
-{
-  'targets': [
-    {
-      'target_name': 'vosk',
-      'sources': [
-         '../src/kaldi_recognizer.cc',
-         '../src/model.cc',
-         '../src/spk_model.cc',
-         '../src/vosk_api.cc',
-         'vosk_wrap.cc',
-      ],
-      'cflags': [
-           '-std=c++11',
-           '-DFST_NO_DYNAMIC_LINKING',
-           '-Wno-deprecated-declarations',
-           '-Wno-sign-compare',
-           '-Wno-unused-local-typedefs',
-           '-Wno-ignored-quaifiers',
-           '-Wno-extra',
-      ],
-      'cflags_cc!' : [
-           '-fno-rtti',
-           '-fno-exceptions',
-      ],
-      'actions': [
-        {
-          'action_name': 'swig',
-          'inputs': [
-            '../src/vosk.i',
-          ],
-          'outputs': [
-            'vosk_wrap.cc',
-          ],
-          'action': ['swig', '-c++', '-javascript', '-o', 'vosk_wrap.cc', '-v8', '-DV8_MAJOR_VERSION=10', '../src/vosk.i']
-        },
-      ],
-      'include_dirs': [
-         '<@(kaldi_root)/src',
-         '<@(kaldi_root)/tools/openfst/include',
-         '../src',
-      ],
-      'link_settings': {
-         'libraries': [
-             '<@(kaldi_root)/src/online2/kaldi-online2.a',
-             '<@(kaldi_root)/src/decoder/kaldi-decoder.a',
-             '<@(kaldi_root)/src/ivector/kaldi-ivector.a',
-             '<@(kaldi_root)/src/gmm/kaldi-gmm.a',
-             '<@(kaldi_root)/src/nnet3/kaldi-nnet3.a',
-             '<@(kaldi_root)/src/tree/kaldi-tree.a',
-             '<@(kaldi_root)/src/feat/kaldi-feat.a',
-             '<@(kaldi_root)/src/lat/kaldi-lat.a',
-             '<@(kaldi_root)/src/lm/kaldi-lm.a',
-             '<@(kaldi_root)/src/hmm/kaldi-hmm.a',
-             '<@(kaldi_root)/src/transform/kaldi-transform.a',
-             '<@(kaldi_root)/src/cudamatrix/kaldi-cudamatrix.a',
-             '<@(kaldi_root)/src/matrix/kaldi-matrix.a',
-             '<@(kaldi_root)/src/fstext/kaldi-fstext.a',
-             '<@(kaldi_root)/src/util/kaldi-util.a',
-             '<@(kaldi_root)/src/base/kaldi-base.a',
-             '<@(kaldi_root)/tools/openfst/lib/libfst.a',
-             '<@(kaldi_root)/tools/openfst/lib/libfstngram.a',
-             '<@(kaldi_root)/tools/OpenBLAS/libopenblas.a',
-          ],
-         'library_dirs': [
-               '/usr/lib',
-          ],
-      },
-    }
-  ]
-}

nodejs/demo/test_ffmpeg.js

@@ -0,0 +1,33 @@
+var vosk = require('..')
+
+const fs = require("fs");
+const { spawn } = require("child_process");
+
+MODEL_PATH = "model"
+FILE_NAME = "test.wav"
+SAMPLE_RATE = 16000
+BUFFER_SIZE = 4000
+
+if (!fs.existsSync(MODEL_PATH)) {
+    console.log("Please download the model from https://alphacephei.com/vosk/models and unpack as " + MODEL_PATH + " in the current folder.")
+    process.exit()
+}
+
+if (process.argv.length > 2)
+    FILE_NAME = process.argv[2]
+
+vosk.setLogLevel(0);
+const model = new vosk.Model(MODEL_PATH);
+const rec = new vosk.Recognizer({model: model, sampleRate: SAMPLE_RATE});
+
+const ffmpeg_run = spawn('ffmpeg', ['-loglevel', 'quiet', '-i', FILE_NAME,
+                         '-ar', String(SAMPLE_RATE) , '-ac', '1',
+                         '-f', 's16le', '-bufsize', String(BUFFER_SIZE) , '-']);
+
+ffmpeg_run.stdout.on('data', (stdout) => {
+    if (rec.acceptWaveform(stdout))
+        console.log(rec.result());
+    else
+        console.log(rec.partialResult());
+    console.log(rec.finalResult());
+});

nodejs/demo/test_microphone.js

@@ -0,0 +1,39 @@
+var vosk = require('..')
+
+const fs = require("fs");
+var mic = require("mic");
+
+MODEL_PATH = "model"
+SAMPLE_RATE = 16000
+
+if (!fs.existsSync(MODEL_PATH)) {
+    console.log("Please download the model from https://alphacephei.com/vosk/models and unpack as " + MODEL_PATH + " in the current folder.")
+    process.exit()
+}
+
+vosk.setLogLevel(0);
+const model = new vosk.Model(MODEL_PATH);
+const rec = new vosk.Recognizer({model: model, sampleRate: SAMPLE_RATE});
+
+var micInstance = mic({
+    rate: String(SAMPLE_RATE),
+    channels: '1',
+    debug: false
+});
+
+var micInputStream = micInstance.getAudioStream();
+micInstance.start();
+
+micInputStream.on('data', data => {
+    if (rec.acceptWaveform(data))
+        console.log(rec.result());
+    else
+        console.log(rec.partialResult());
+});
+
+process.on('SIGINT', function() {
+    console.log(rec.finalResult());
+    console.log("\nDone");
+    rec.free();
+    model.free();
+});

nodejs/demo/test_simple.js

@@ -0,0 +1,45 @@
+var vosk = require('..')
+
+const fs = require("fs");
+const { Readable } = require("stream");
+const wav = require("wav");
+
+MODEL_PATH = "model"
+FILE_NAME = "test.wav"
+
+if (!fs.existsSync(MODEL_PATH)) {
+    console.log("Please download the model from https://alphacephei.com/vosk/models and unpack as " + MODEL_PATH + " in the current folder.")
+    process.exit()
+}
+
+if (process.argv.length > 2)
+    FILE_NAME = process.argv[2]
+
+vosk.setLogLevel(0);
+const model = new vosk.Model(MODEL_PATH);
+
+const wfReader = new wav.Reader();
+const wfReadable = new Readable().wrap(wfReader);
+
+wfReader.on('format', async ({ audioFormat, sampleRate, channels }) => {
+    if (audioFormat != 1 || channels != 1) {
+        console.error("Audio file must be WAV format mono PCM.");
+        process.exit(1);
+    }
+    const rec = new vosk.Recognizer({model: model, sampleRate: sampleRate});
+    rec.setMaxAlternatives(10);
+    rec.setWords(true);
+    for await (const data of wfReadable) {
+        const end_of_speech = rec.acceptWaveform(data);
+        if (end_of_speech) {
+              console.log(JSON.stringify(rec.result(), null, 4));
+        }
+    }
+    console.log(JSON.stringify(rec.finalResult(rec), null, 4));
+    rec.free();
+});
+
+fs.createReadStream(FILE_NAME, {'highWaterMark': 4096}).pipe(wfReader).on('finish', 
+    function (err) {
+        model.free();
+});

nodejs/demo/test_simple_async.js

@@ -0,0 +1,46 @@
+var vosk = require('..')
+
+const async = require("async");
+const fs = require("fs");
+const { Readable } = require("stream");
+const wav = require("wav");
+
+MODEL_PATH = "model"
+
+if (!fs.existsSync(MODEL_PATH)) {
+    console.log("Please download the model from https://alphacephei.com/vosk/models and unpack as " + MODEL_PATH + " in the current folder.")
+    process.exit()
+}
+
+// Process file 4 times in parallel with a single model
+files = Array(10).fill("test.wav")
+const model = new vosk.Model(MODEL_PATH)
+
+async.filter(files, function(filePath, callback) {
+    const wfReader = new wav.Reader();
+    const wfReadable = new Readable().wrap(wfReader);
+
+    wfReader.on('format', async ({ audioFormat, sampleRate, channels }) => {
+        const rec = new vosk.Recognizer({model: model, sampleRate: sampleRate});
+        if (audioFormat != 1 || channels != 1) {
+            console.error("Audio file must be WAV format mono PCM.");
+            process.exit(1);
+        }
+        for await (const data of wfReadable) {
+            const end_of_speech = await rec.acceptWaveformAsync(data);
+            if (end_of_speech) {
+                  console.log(rec.result());
+            }
+        }
+        console.log(rec.finalResult(rec));
+        rec.free();
+        // Signal we are done without errors
+        callback(null, true);
+    });
+
+    fs.createReadStream(filePath, {'highWaterMark': 4096}).pipe(wfReader);
+
+}, function(err, results) {
+    model.free();
+    console.log("Done!!!!!");
+});

nodejs/demo/test_speaker.js

@@ -0,0 +1,53 @@
+const vosk = require('..');
+
+const fs = require("fs");
+const { Readable } = require("stream");
+const wav = require("wav");
+
+MODEL_PATH = "model"
+SPEAKER_MODEL_PATH = "model-spk"
+FILE_NAME = "test.wav"
+
+if (!fs.existsSync(MODEL_PATH)) {
+    console.log("Please download the model from https://alphacephei.com/vosk/models and unpack as " + MODEL_PATH + " in the current folder.")
+    process.exit()
+}
+
+if (!fs.existsSync(SPEAKER_MODEL_PATH)) {
+    console.log("Please download the speaker model from https://alphacephei.com/vosk/models and unpack as " + SPEAKER_MODEL_PATH + " in the current folder.")
+    process.exit()
+}
+
+if (process.argv.length > 2)
+    FILE_NAME = process.argv[2]
+
+const model = new vosk.Model(MODEL_PATH);
+const speakerModel = new vosk.SpeakerModel(SPEAKER_MODEL_PATH);
+
+const wfReader = new wav.Reader();
+const wfReadable = new Readable().wrap(wfReader);
+
+wfReader.on('format', async ({ audioFormat, sampleRate, channels }) => {
+    if (audioFormat != 1 || channels != 1) {
+        console.error('Audio file must be WAV format mono PCM.');
+        process.exit(1);
+    }
+//    const rec = new vosk.Recognizer({ model: model,
+//                                      speakerModel: speakerModel,
+//                                      sampleRate: sampleRate });
+    const rec = new vosk.Recognizer({model: model, sampleRate: sampleRate});
+    rec.setSpkModel(speakerModel);
+    for await (const data of wfReadable) {
+        const end_of_speech = rec.acceptWaveform(data);
+        if (end_of_speech) {
+             console.log(rec.finalResult());
+        }
+    }
+    console.log(rec.finalResult());
+    rec.free();
+});
+
+fs.createReadStream(FILE_NAME, { highWaterMark: 4096 }).pipe(wfReader).on('finish', function (err) {
+    model.free();
+    speakerModel.free();
+});

nodejs/demo/test_srt.js

@@ -0,0 +1,84 @@
+var vosk = require('..')
+
+const fs = require("fs");
+const { spawn } = require("child_process");
+const { stringifySync } = require('subtitle')
+
+MODEL_PATH = "model"
+FILE_NAME = "test.wav"
+SAMPLE_RATE = 16000 
+BUFFER_SIZE = 4000
+
+if (!fs.existsSync(MODEL_PATH)) {
+    console.log("Please download the model from https://alphacephei.com/vosk/models and unpack as " + MODEL_PATH + " in the current folder.")
+    process.exit()
+}
+
+if (process.argv.length > 2)
+    FILE_NAME = process.argv[2]
+
+vosk.setLogLevel(-1);
+const model = new vosk.Model(MODEL_PATH);
+const rec = new vosk.Recognizer({model: model, sampleRate: SAMPLE_RATE});
+rec.setWords(true);
+
+const ffmpeg_run = spawn('ffmpeg', ['-loglevel', 'quiet', '-i', FILE_NAME,
+                         '-ar', String(SAMPLE_RATE) , '-ac', '1',
+                         '-f', 's16le', '-bufsize', String(BUFFER_SIZE), '-']);
+
+WORDS_PER_LINE = 7
+const subs = []
+const results = []
+ffmpeg_run.stdout.on('data', (stdout) => {
+    if (rec.acceptWaveform(stdout))
+        results.push(rec.result());
+    results.push(rec.finalResult());
+});
+
+ffmpeg_run.on('exit', code => {
+    rec.free();
+    model.free();
+    results.forEach(element =>{
+        if (!element.hasOwnProperty('result'))
+            return;
+        const words = element.result;
+        if (words.length == 1) {
+            subs.push({
+                type: 'cue',
+                data: {
+                start: words[0].start,
+                end: words[0].end,
+                text: words[0].word
+                }
+            });
+            return;
+        }
+        var start_index = 0;
+        var text = words[0].word + " ";
+        for (let i = 1; i < words.length; i++) {
+            text += words[i].word + " ";
+            if (i % WORDS_PER_LINE == 0) {
+                subs.push({
+                    type: 'cue',
+                    data: {
+                      start: words[start_index].start,
+                      end: words[i].end,
+                      text: text.slice(0, text.length-1)
+                    }
+                });
+                start_index = i;
+                text = "";
+            }
+        }
+        if (start_index != words.length - 1)
+            subs.push({
+                type: 'cue',
+                data: {
+                  start: words[start_index].start,
+                  end: words[words.length-1].end,
+                  text: text
+                }
+            });
+    });
+    console.log(stringifySync(subs, {format: "SRT"}));
+});

nodejs/example/test-speaker.js

@@ -1,47 +0,0 @@
-#!/usr/bin/env node
-
-const fs = require("fs");
-const { Readable } = require("stream");
-const wav = require("wav");
-
-const { Model, KaldiRecognizer, SpkModel } = require("..");
-
-try {
-    fs.accessSync("model", fs.constants.R_OK);
-} catch(err) {
-    console.error("Please download the model from https://github.com/alphacep/kaldi-android-demo/releases and unpack as 'model' in the current folder.");
-    process.exit(1);
-}
-
-try {
-    fs.accessSync("model-spk", fs.constants.R_OK);
-} catch(err) {
-    console.error("Please download the speaker model from https://github.com/alphacep/kaldi-android-demo/releases and unpack as 'model-spk' in the current folder.");
-    process.exit(1);
-}
-
-
-const wfStream = fs.createReadStream("test.wav", {'highWaterMark': 4096});
-const wfReader = new wav.Reader();
-
-const model = new Model("model");
-const spkModel = new SpkModel("model-spk");
-
-wfReader.on('format', async ({ audioFormat, sampleRate, channels }) => {
-    if (audioFormat != 1 || channels != 1) {
-        console.error("Audio file must be WAV format mono PCM.");
-        process.exit(1);
-    }
-    const rec = new KaldiRecognizer(model, spkModel, sampleRate);
-    for await (const data of new Readable().wrap(wfReader)) {
-        const result = await rec.AcceptWaveform(data);
-        if (result != 0) {
-           console.log(await rec.Result());
-        } else {
-           console.log(await rec.PartialResult());
-        }
-    }
-    console.log(await rec.FinalResult());
-});
-
-wfStream.pipe(wfReader);

nodejs/example/test.js

@@ -1,38 +0,0 @@
-#!/usr/bin/env node
-
-const fs = require("fs");
-const { Readable } = require("stream");
-const wav = require("wav");
-
-const { Model, KaldiRecognizer } = require("..");
-
-try {
-    fs.accessSync("model", fs.constants.R_OK);
-} catch(err) {
-    console.error("Please download the model from https://github.com/alphacep/kaldi-android-demo/releases and unpack as 'model' in the current folder.");
-    process.exit(1);
-}
-
-const wfStream = fs.createReadStream("test.wav", {'highWaterMark': 4096});
-const wfReader = new wav.Reader();
-
-const model = new Model("model");
-
-wfReader.on('format', async ({ audioFormat, sampleRate, channels }) => {
-    if (audioFormat != 1 || channels != 1) {
-        console.error("Audio file must be WAV format mono PCM.");
-        process.exit(1);
-    }
-    const rec = new KaldiRecognizer(model, sampleRate);
-    for await (const data of new Readable().wrap(wfReader)) {
-        const result = await rec.AcceptWaveform(data);
-        if (result != 0) {
-           console.log(await rec.Result());
-        } else {
-           console.log(await rec.PartialResult());
-        }
-    }
-    console.log(await rec.FinalResult());
-});
-
-wfStream.pipe(wfReader);

nodejs/index.js

@@ -1,3 +1,427 @@
-const voskNativeModule = require('./build/Release/vosk.node');
+// @ts-check
+'use strict'
 
-module.exports = voskNativeModule;
+/**
+ * @module vosk
+ */
+
+const os = require('os');
+const path = require('path');
+/** @type {import('ffi-napi')} */
+const ffi = require('ffi-napi');
+/** @type {import('ref-napi')} */
+const ref = require('ref-napi');
+
+const vosk_model = ref.types.void;
+const vosk_model_ptr = ref.refType(vosk_model);
+const vosk_spk_model = ref.types.void;
+const vosk_spk_model_ptr = ref.refType(vosk_spk_model);
+const vosk_recognizer = ref.types.void;
+const vosk_recognizer_ptr = ref.refType(vosk_recognizer);
+
+/**
+ * @typedef {Object} WordResult
+ * @property {number} conf The confidence rate in the detection. 0 For unlikely, and 1 for totally accurate.
+ * @property {number} start The start of the timeframe when the word is pronounced in seconds
+ * @property {number} end The end of the timeframe when the word is pronounced in seconds
+ * @property {string} word The word detected
+ */
+
+/**
+ * @typedef {Object} RecognitionResults
+ * @property {WordResult[]} result Details about the words that have been detected
+ * @property {string} text The complete sentence that have been detected
+ */
+
+/**
+ * @typedef {Object} SpeakerResults
+ * @property {number[]} spk A floating vector representing speaker identity. It is usually about 128 numbers which uniquely represent speaker voice.
+ * @property {number} spk_frames The number of frames used to extract speaker vector. The more frames you have the more reliable is speaker vector.
+ */
+
+/**
+ * @typedef {Object} BaseRecognizerParam
+ * @property {Model} model The language model to be used 
+ * @property {number} sampleRate The sample rate. Most models are trained at 16kHz
+ */
+
+/**
+ * @typedef {Object} GrammarRecognizerParam
+ * @property {string[]} grammar The list of sentences to be recognized.
+ */
+
+/**
+ * @typedef {Object} SpeakerRecognizerParam
+ * @property {SpeakerModel} speakerModel The SpeakerModel that will enable speaker identification
+ */
+
+/**
+ * @template {SpeakerRecognizerParam | GrammarRecognizerParam} T
+ * @typedef {T extends SpeakerRecognizerParam ? SpeakerResults & RecognitionResults : RecognitionResults} Result
+ */
+
+/**
+ * @typedef {Object} PartialResults
+ * @property {string} partial The partial sentence that have been detected until now
+ */
+
+/** @typedef {string[]} Grammar The list of strings to be recognized */
+
+let soname;
+if (os.platform() == 'win32') {
+    // Update path to load dependent dlls
+    let currentPath = process.env.Path;
+    let dllDirectory = path.resolve(path.join(__dirname, "lib", "win-x86_64"));
+    process.env.Path = currentPath + path.delimiter + dllDirectory;
+
+    soname = path.join(__dirname, "lib", "win-x86_64", "libvosk.dll")
+} else if (os.platform() == 'darwin') {
+    soname = path.join(__dirname, "lib", "osx-x86_64", "libvosk.dylib")
+} else {
+    soname = path.join(__dirname, "lib", "linux-x86_64", "libvosk.so")
+}
+
+const libvosk = ffi.Library(soname, {
+    'vosk_set_log_level': ['void', ['int']],
+    'vosk_model_new': [vosk_model_ptr, ['string']],
+    'vosk_model_free': ['void', [vosk_model_ptr]],
+    'vosk_spk_model_new': [vosk_spk_model_ptr, ['string']],
+    'vosk_spk_model_free': ['void', [vosk_spk_model_ptr]],
+    'vosk_recognizer_new': [vosk_recognizer_ptr, [vosk_model_ptr, 'float']],
+    'vosk_recognizer_new_spk': [vosk_recognizer_ptr, [vosk_model_ptr, 'float', vosk_spk_model_ptr]],
+    'vosk_recognizer_new_grm': [vosk_recognizer_ptr, [vosk_model_ptr, 'float', 'string']],
+    'vosk_recognizer_free': ['void', [vosk_recognizer_ptr]],
+    'vosk_recognizer_set_max_alternatives': ['void', [vosk_recognizer_ptr, 'int']],
+    'vosk_recognizer_set_words': ['void', [vosk_recognizer_ptr, 'bool']],
+    'vosk_recognizer_set_spk_model': ['void', [vosk_recognizer_ptr, vosk_spk_model_ptr]],
+    'vosk_recognizer_accept_waveform': ['bool', [vosk_recognizer_ptr, 'pointer', 'int']],
+    'vosk_recognizer_result': ['string', [vosk_recognizer_ptr]],
+    'vosk_recognizer_final_result': ['string', [vosk_recognizer_ptr]],
+    'vosk_recognizer_partial_result': ['string', [vosk_recognizer_ptr]],
+    'vosk_recognizer_reset': ['void', [vosk_recognizer_ptr]],
+});
+
+/**
+ * Set log level for Kaldi messages
+ * @param {number} level The higher, the more verbose. 0 for infos and errors. Less than 0 for silence. 
+ */
+function setLogLevel(level) {
+    libvosk.vosk_set_log_level(level);
+}
+
+/**
+ * Build a Model from a model file.
+ * @see models [models](https://alphacephei.com/vosk/models)
+ */
+class Model {
+    /**
+     * Build a Model to be used with the voice recognition. Each language should have it's own Model
+     * for the speech recognition to work.
+     * @param {string} modelPath The abstract pathname to the model
+     * @see models [models](https://alphacephei.com/vosk/models)
+     */
+    constructor(modelPath) {
+        /**
+         * Store the handle.
+         * For internal use only
+         * @type {unknown}
+         */
+        this.handle = libvosk.vosk_model_new(modelPath);
+    }
+
+    /**
+     * 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.
+     */
+    free() {
+        libvosk.vosk_model_free(this.handle);
+    }
+}
+
+/**
+ * Build a Speaker Model from a speaker model file.
+ * The Speaker Model enables speaker identification.
+ * @see models [models](https://alphacephei.com/vosk/models)
+ */
+class SpeakerModel {
+    /**
+     * Loads speaker model data from the file and returns the model object
+     *
+     * @param {string} modelPath the path of the model on the filesystem
+     * @see models [models](https://alphacephei.com/vosk/models)
+     */
+    constructor(modelPath) {
+        /**
+         * Store the handle.
+         * For internal use only
+         * @type {unknown}
+         */
+        this.handle = libvosk.vosk_spk_model_new(modelPath);
+    }
+
+    /**
+     * 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.
+     */
+    free() {
+        libvosk.vosk_spk_model_free(this.handle);
+    }
+}
+
+/**
+ * Helper to narrow down type while using `hasOwnProperty`.
+ * @see hasOwnProperty [typescript issue](https://fettblog.eu/typescript-hasownproperty/)
+ * @template {Object} Obj
+ * @template {PropertyKey} Key
+ * @param {Obj} obj 
+ * @param {Key} prop 
+ * @returns {obj is Obj & Record<Key, unknown>}
+ */
+function hasOwnProperty(obj, prop) {
+    return obj.hasOwnProperty(prop)
+}
+
+/**
+ * @template T
+ * @template U
+ * @typedef {{ [P in Exclude<keyof T, keyof U>]?: never }} Without
+ */
+
+/**
+ * @template T
+ * @template U
+ * @typedef {(T | U) extends object ? (Without<T, U> & U) | (Without<U, T> & T) : T | U} XOR
+ */
+
+/**
+ * Create a Recognizer that will be able to transform audio streams into text using a Model.
+ * @template {XOR<SpeakerRecognizerParam, Partial<GrammarRecognizerParam>>} T extra parameter
+ * @see Model
+ */
+class Recognizer {
+    /**
+     * Create a Recognizer that will handle speech to text recognition.
+     * @constructor
+     * @param {T & BaseRecognizerParam} param The Recognizer parameters 
+     *
+     *  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.
+     */
+    constructor(param) {
+        const { model, sampleRate } = param
+        // Prevent the user to receive unpredictable results
+        if (hasOwnProperty(param, 'speakerModel') && hasOwnProperty(param, 'grammar')) {
+            throw new Error('grammar and speakerModel cannot be used together for now.')
+        }
+        /**
+         * Store the handle.
+         * For internal use only
+         * @type {unknown}
+         */
+        this.handle = hasOwnProperty(param, 'speakerModel')
+            ? libvosk.vosk_recognizer_new_spk(model.handle, sampleRate, param.speakerModel.handle)
+            : hasOwnProperty(param, 'grammar')
+                ? libvosk.vosk_recognizer_new_grm(model.handle, sampleRate, JSON.stringify(param.grammar))
+                : libvosk.vosk_recognizer_new(model.handle, sampleRate);
+    }
+
+    /**
+     * 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.
+     */
+    free() {
+        libvosk.vosk_recognizer_free(this.handle);
+    }
+
+    /** 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
+     */
+    setMaxAlternatives(max_alternatives) {
+        libvosk.vosk_recognizer_set_max_alternatives(this.handle, max_alternatives);
+    }
+
+    /** Configures recognizer to output words with times
+     *
+     * <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
+     */
+    setWords(words) {
+        libvosk.vosk_recognizer_set_words(this.handle, words);
+    }
+
+    /** Adds speaker recognition model to already created recognizer. Helps to initialize
+     * speaker recognition for grammar-based recognizer.
+     *
+     * @param spk_model Speaker recognition model
+     */
+    setSpkModel(spk_model) {
+        libvosk.vosk_recognizer_set_spk_model(this.handle, spk_model.handle);
+    }
+
+    /** 
+     * Accept voice data
+     *
+     * accept and process new chunk of voice data
+     *
+     * @param {Buffer} data audio data in PCM 16-bit mono format
+     * @returns true if silence is occured and you can retrieve a new utterance with result method
+     */
+    acceptWaveform(data) {
+        return libvosk.vosk_recognizer_accept_waveform(this.handle, data, data.length);
+    };
+
+    /** 
+     * Accept voice data
+     *
+     * accept and process new chunk of voice data
+     *
+     * @param {Buffer} data audio data in PCM 16-bit mono format
+     * @returns true if silence is occured and you can retrieve a new utterance with result method
+     */
+    acceptWaveformAsync(data) {
+        return new Promise((resolve, reject) => {
+            libvosk.vosk_recognizer_accept_waveform.async(this.handle, data, data.length, function(err, result) {
+                if (err) {
+                    reject(err);
+                } else {
+                    resolve(result);
+                }
+            });
+        });
+    };
+
+    /** Returns speech recognition result in a string
+     *
+     * @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>
+     * {
+     *   "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"
+     *     }],
+     *   "text" : "what zero zero zero one"
+     *  }
+     * </pre>
+     */
+    resultString() {
+        return libvosk.vosk_recognizer_result(this.handle);
+    };
+
+    /**
+     * Returns speech recognition results
+     * @returns {Result<T>} The results
+     */
+    result() {
+        return JSON.parse(libvosk.vosk_recognizer_result(this.handle));
+    };
+
+    /**
+     * speech recognition text which is not yet finalized.
+     * result may change as recognizer process more data.
+     * 
+     * @returns {PartialResults} The partial results
+     */
+    partialResult() {
+        return JSON.parse(libvosk.vosk_recognizer_partial_result(this.handle));
+    };
+
+    /**
+     * 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 {Result<T>} speech result.
+     */
+    finalResult() {
+        return JSON.parse(libvosk.vosk_recognizer_final_result(this.handle));
+    };
+
+    /**
+     *
+     * Resets current results so the recognition can continue from scratch 
+     */
+    reset() {
+        libvosk.vosk_recognizer_reset(this.handle);
+    }
+}
+
+exports.setLogLevel = setLogLevel
+exports.Model = Model
+exports.SpeakerModel = SpeakerModel
+exports.Recognizer = Recognizer

nodejs/package.json

@@ -1,7 +1,7 @@
 {
   "name": "vosk",
-  "version": "0.3.8",
-  "description": "Node binding for continuous voice recoginition through pocketsphinx.",
+  "version": "0.3.32",
+  "description": "Node binding for continuous offline voice recoginition with Vosk library.",
   "repository": {
     "type": "git",
     "url": "git://github.com/alphacep/vosk-api.git"
@@ -15,10 +15,13 @@
   "author": "Alpha Cephei Inc.",
   "license": "Apache-2.0",
   "engines": {
-    "node": ">= 10.x.x"
+    "node": ">= 12.x.x"
   },
   "dependencies": {
-    "node-gyp": "^5.1.1",
+    "async": "^3.2.0",
+    "ffi-napi": "^4.0.3",
+    "mic": "^2.1.2",
+    "ref-napi": ">=2.0.0",
     "wav": "^1.0.2"
   }
 }

python/README.md

@@ -1,3 +1,23 @@
-Python module for vosk-api
+This is a Python module for Vosk.
 
-See for details https://github.com/alphacep/vosk-api
+Vosk is an offline open source speech recognition toolkit. It enables
+speech recognition models for 17 languages and dialects - English, Indian
+English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish,
+Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi, Filipino.
+
+Vosk models are small (50 Mb) but provide continuous large vocabulary
+transcription, zero-latency response with streaming API, reconfigurable
+vocabulary and speaker identification.
+
+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). See also our project on
+[Github](https://github.com/alphacep/vosk-api).

python/example/test_alternatives.py

@@ -0,0 +1,34 @@
+#!/usr/bin/env python3
+
+from vosk import Model, KaldiRecognizer, SetLogLevel
+import sys
+import os
+import wave
+import json
+
+SetLogLevel(0)
+
+if not os.path.exists("model"):
+    print ("Please download the model from https://alphacephei.com/vosk/models and unpack as 'model' in the current folder.")
+    exit (1)
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print ("Audio file must be WAV format mono PCM.")
+    exit (1)
+
+model = Model("model")
+rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetMaxAlternatives(10)
+rec.SetWords(True)
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(json.loads(rec.Result()))
+    else:
+        print(json.loads(rec.PartialResult()))
+
+print(json.loads(rec.FinalResult()))

python/example/test_ffmpeg.py

@@ -9,7 +9,7 @@ import subprocess
 SetLogLevel(0)
 
 if not os.path.exists("model"):
-    print ("Please download the model from https://github.com/alphacep/vosk-api/blob/master/doc/models.md and unpack as 'model' in the current folder.")
+    print ("Please download the model from https://alphacephei.com/vosk/models and unpack as 'model' in the current folder.")
     exit (1)
 
 sample_rate=16000

python/example/test_microphone.py

@@ -1,28 +1,89 @@
 #!/usr/bin/env python3
 
-from vosk import Model, KaldiRecognizer
+import argparse
 import os
+import queue
+import sounddevice as sd
+import vosk
+import sys
 
-if not os.path.exists("model"):
-    print ("Please download the model from https://github.com/alphacep/vosk-api/blob/master/doc/models.md and unpack as 'model' in the current folder.")
-    exit (1)
+q = queue.Queue()
 
-import pyaudio
+def int_or_str(text):
+    """Helper function for argument parsing."""
+    try:
+        return int(text)
+    except ValueError:
+        return text
 
-model = Model("model")
-rec = KaldiRecognizer(model, 16000)
+def callback(indata, frames, time, status):
+    """This is called (from a separate thread) for each audio block."""
+    if status:
+        print(status, file=sys.stderr)
+    q.put(bytes(indata))
 
-p = pyaudio.PyAudio()
-stream = p.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True, frames_per_buffer=8000)
-stream.start_stream()
+parser = argparse.ArgumentParser(add_help=False)
+parser.add_argument(
+    '-l', '--list-devices', action='store_true',
+    help='show list of audio devices and exit')
+args, remaining = parser.parse_known_args()
+if args.list_devices:
+    print(sd.query_devices())
+    parser.exit(0)
+parser = argparse.ArgumentParser(
+    description=__doc__,
+    formatter_class=argparse.RawDescriptionHelpFormatter,
+    parents=[parser])
+parser.add_argument(
+    '-f', '--filename', type=str, metavar='FILENAME',
+    help='audio file to store recording to')
+parser.add_argument(
+    '-m', '--model', type=str, metavar='MODEL_PATH',
+    help='Path to the model')
+parser.add_argument(
+    '-d', '--device', type=int_or_str,
+    help='input device (numeric ID or substring)')
+parser.add_argument(
+    '-r', '--samplerate', type=int, help='sampling rate')
+args = parser.parse_args(remaining)
 
-while True:
-    data = stream.read(4000)
-    if len(data) == 0:
-        break
-    if rec.AcceptWaveform(data):
-        print(rec.Result())
+try:
+    if args.model is None:
+        args.model = "model"
+    if not os.path.exists(args.model):
+        print ("Please download a model for your language from https://alphacephei.com/vosk/models")
+        print ("and unpack as 'model' in the current folder.")
+        parser.exit(0)
+    if args.samplerate is None:
+        device_info = sd.query_devices(args.device, 'input')
+        # soundfile expects an int, sounddevice provides a float:
+        args.samplerate = int(device_info['default_samplerate'])
+
+    model = vosk.Model(args.model)
+
+    if args.filename:
+        dump_fn = open(args.filename, "wb")
     else:
-        print(rec.PartialResult())
+        dump_fn = None
 
-print(rec.FinalResult())
+    with sd.RawInputStream(samplerate=args.samplerate, blocksize = 8000, device=args.device, dtype='int16',
+                            channels=1, callback=callback):
+            print('#' * 80)
+            print('Press Ctrl+C to stop the recording')
+            print('#' * 80)
+
+            rec = vosk.KaldiRecognizer(model, args.samplerate)
+            while True:
+                data = q.get()
+                if rec.AcceptWaveform(data):
+                    print(rec.Result())
+                else:
+                    print(rec.PartialResult())
+                if dump_fn is not None:
+                    dump_fn.write(data)
+
+except KeyboardInterrupt:
+    print('\nDone')
+    parser.exit(0)
+except Exception as e:
+    parser.exit(type(e).__name__ + ': ' + str(e))

python/example/test_reset.py

@@ -0,0 +1,37 @@
+#!/usr/bin/env python3
+
+from vosk import Model, KaldiRecognizer, SetLogLevel
+import sys
+import os
+import wave
+import json
+
+SetLogLevel(0)
+
+if not os.path.exists("model"):
+    print ("Please download the model from https://alphacephei.com/vosk/models and unpack as 'model' in the current folder.")
+    exit (1)
+
+wf = wave.open(sys.argv[1], "rb")
+if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE":
+    print ("Audio file must be WAV format mono PCM.")
+    exit (1)
+
+model = Model("model")
+rec = KaldiRecognizer(model, wf.getframerate())
+
+while True:
+    data = wf.readframes(4000)
+    if len(data) == 0:
+        break
+    if rec.AcceptWaveform(data):
+        print(rec.Result())
+        break
+
+    else:
+        jres = json.loads(rec.PartialResult())
+        print(jres)
+
+        if jres['partial'] == "one zero zero zero":
+            print("We can reset recognizer here and start over")
+            rec.Reset();

python/example/test_simple.py

@@ -8,7 +8,7 @@ import wave
 SetLogLevel(0)
 
 if not os.path.exists("model"):
-    print ("Please download the model from https://github.com/alphacep/vosk-api/blob/master/doc/models.md and unpack as 'model' in the current folder.")
+    print ("Please download the model from https://alphacephei.com/vosk/models and unpack as 'model' in the current folder.")
     exit (1)
 
 wf = wave.open(sys.argv[1], "rb")
@@ -18,6 +18,7 @@ if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE
 
 model = Model("model")
 rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetWords(True)
 
 while True:
     data = wf.readframes(4000)

python/example/test_speaker.py

@@ -11,11 +11,11 @@ model_path = "model"
 spk_model_path = "model-spk"
 
 if not os.path.exists(model_path):
-    print ("Please download the model from https://github.com/alphacep/vosk-api/blob/master/doc/models.md and unpack as {} in the current folder.".format(model_path))
+    print ("Please download the model from https://alphacephei.com/vosk/models and unpack as {} in the current folder.".format(model_path))
     exit (1)
 
 if not os.path.exists(spk_model_path):
-    print ("Please download the speaker model from https://github.com/alphacep/vosk-api/blob/master/doc/models.md and unpack as {} in the current folder.".format(spk_model_path))
+    print ("Please download the speaker model from https://alphacephei.com/vosk/models and unpack as {} in the current folder.".format(spk_model_path))
     exit (1)
 
 wf = wave.open(sys.argv[1], "rb")
@@ -26,11 +26,13 @@ if wf.getnchannels() != 1 or wf.getsampwidth() != 2 or wf.getcomptype() != "NONE
 # Large vocabulary free form recognition
 model = Model(model_path)
 spk_model = SpkModel(spk_model_path)
-rec = KaldiRecognizer(model, spk_model, wf.getframerate())
+#rec = KaldiRecognizer(model, wf.getframerate(), spk_model)
+rec = KaldiRecognizer(model, wf.getframerate())
+rec.SetSpkModel(spk_model)
 
 # We compare speakers with cosine distance. We can keep one or several fingerprints for the speaker in a database
 # to distingusih among users.
-spk_sig = [4.658117, 1.277387, 3.346158, -1.473036, -2.15727, 2.461757, 3.76756, -1.241252, 2.333765, 0.642588, -2.848165, 1.229534, 3.907015, 1.726496, -1.188692, 1.16322, -0.668811, -0.623309, 4.628018, 0.407197, 0.089955, 0.920438, 1.47237, -0.311365, -0.437051, -0.531738, -1.591781, 3.095415, 0.439524, -0.274787, 4.03165, 2.665864, 4.815553, 1.581063, 1.078242, 5.017717, -0.089395, -3.123428, 5.34038, 0.456982, 2.465727, 2.131833, 4.056272, 1.178392, -2.075712, -1.568503, 0.847139, 0.409214, 1.84727, 0.986758, 4.222116, 2.235512, 1.369377, 4.283126, 2.278125, -1.467577, -0.999971, 3.070041, 1.462214, 0.423204, 2.143578, 0.567174, -2.294655, 1.864723, 4.307356, 2.610872, -1.238721, 0.551861, 2.861954, 0.59613, -0.715396, -1.395357, 2.706177, -2.004444, 2.055255, 0.458283, 1.231968, 3.48234, 2.993858, 0.402819, 0.940885, 0.360162, -2.173674, -2.504609, 0.329541, 3.653913, 3.638025, -1.406409, 2.14059, 1.662765, -0.991323, 0.770921, 0.010094, 3.775469, 1.847511, 2.074432, -1.928593, 0.807414, 2.964505, 0.128597, 1.297962, 2.645227, 0.136405, -2.543087, 0.932246, 2.405783, -2.122267, 3.044013, 0.486728, 4.395338, 0.474267, 0.781297, 1.694144, -0.831078, -0.462362, -0.964715, 3.187863, 6.008708, 1.725954, 3.667886, -1.467623, 3.370667, 2.72555, -0.796541, 2.416543, 0.675401, -0.737634, -1.709676]
+spk_sig = [-1.110417,0.09703002,1.35658,0.7798632,-0.305457,-0.339204,0.6186931,-0.4521213,0.3982236,-0.004530723,0.7651616,0.6500852,-0.6664245,0.1361499,0.1358056,-0.2887807,-0.1280468,-0.8208137,-1.620276,-0.4628615,0.7870904,-0.105754,0.9739769,-0.3258137,-0.7322628,-0.6212429,-0.5531687,-0.7796484,0.7035915,1.056094,-0.4941756,-0.6521456,-0.2238328,-0.003737517,0.2165709,1.200186,-0.7737719,0.492015,1.16058,0.6135428,-0.7183084,0.3153541,0.3458071,-1.418189,-0.9624157,0.4168292,-1.627305,0.2742135,-0.6166027,0.1962581,-0.6406527,0.4372789,-0.4296024,0.4898657,-0.9531326,-0.2945702,0.7879696,-1.517101,-0.9344181,-0.5049928,-0.005040941,-0.4637912,0.8223695,-1.079849,0.8871287,-0.9732434,-0.5548235,1.879138,-1.452064,-0.1975368,1.55047,0.5941782,-0.52897,1.368219,0.6782904,1.202505,-0.9256122,-0.9718158,-0.9570228,-0.5563112,-1.19049,-1.167985,2.606804,-2.261825,0.01340385,0.2526799,-1.125458,-1.575991,-0.363153,0.3270262,1.485984,-1.769565,1.541829,0.7293826,0.1743717,-0.4759418,1.523451,-2.487134,-1.824067,-0.626367,0.7448186,-1.425648,0.3524166,-0.9903384,3.339342,0.4563958,-0.2876643,1.521635,0.9508078,-0.1398541,0.3867955,-0.7550205,0.6568405,0.09419366,-1.583935,1.306094,-0.3501927,0.1794427,-0.3768163,0.9683866,-0.2442541,-1.696921,-1.8056,-0.6803037,-1.842043,0.3069353,0.9070363,-0.486526]
 
 def cosine_dist(x, y):
     nx = np.array(x)
@@ -44,10 +46,15 @@ while True:
     if rec.AcceptWaveform(data):
         res = json.loads(rec.Result())
         print ("Text:", res['text'])
-        print ("X-vector:", res['spk'])
-        print ("Speaker distance:", cosine_dist(spk_sig, res['spk']))
+        if 'spk' in res:
+            print ("X-vector:", res['spk'])
+            print ("Speaker distance:", cosine_dist(spk_sig, res['spk']), "based on", res['spk_frames'], "frames")
+
+print ("Note that second distance is not very reliable because utterance is too short. Utterances longer than 4 seconds give better xvector")
 
 res = json.loads(rec.FinalResult())
 print ("Text:", res['text'])
-print ("Speaker distance:", cosine_dist(spk_sig, res['spk']))
+if 'spk' in res:
+   print ("X-vector:", res['spk'])
+   print ("Speaker distance:", cosine_dist(spk_sig, res['spk']), "based on", res['spk_frames'], "frames")