Update version

Some more links
Document model training and model structure
2020-05-03 20:52:43 +02:00 · 2020-05-02 01:37:10 +02:00 · 2020-05-02 01:11:25 +02:00 · 2020-05-01 20:02:46 +02:00 · 2020-05-01 19:02:57 +02:00 · 2020-04-30 15:19:13 +02:00
54 changed files with 1951 additions and 265 deletions
@@ -34,4 +34,4 @@ java/*.cc
 # CSharp
 csharp/gen
 csharp/*.exe
-csharp/*.cc
+csharp/*.c
@@ -1,11 +1,19 @@
 [![Build Status](https://travis-ci.com/alphacep/vosk-api.svg?branch=master)](https://travis-ci.com/alphacep/vosk-api)

-Language bindings for Vosk and Kaldi to access speech recognition from various languages and on various platforms
+[РУС](README.ru.md)

-  * Python on Linux, Windows and RPi
-  * Node
-  * Android
-  * iOS
+[中文](README.zh.md)
+
+Vosk is a speech recognition toolkit. The best things in Vosk are:
+
+  1. Supports 9 languages - English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish, Vietnamese. More to come.
+  1. Works offline, even on lightweight devices - Raspberry Pi, Android, iOS
+  1. Installs with simple `pip3 install vosk`
+  1. Portable per-language models are only 50Mb each, but there are much bigger server models available.
+  1. Provides streaming API for the best user experience (unlike popular speech-recognition python packages)
+  1. There are bindings for different programming languages, too - java/csharp/javascript etc.
+  1. Allows quick reconfiguration of vocabulary for best accuracy.
+  1. Supports speaker identification beside simple speech recognition.

 ## Android build

@@ -16,6 +24,12 @@ gradle build

 Please note that medium blog post about 64-bit is not relevant anymore, the script builds x86, arm64 and armv7 libraries automatically without any modifications.

+For example of Android application using Vosk-API check https://github.com/alphacep/kaldi-android-demo project
+
+## iOS build
+
+Available on request. Drop as a mail at [contact@alphacephei.com](mailto:contact@alphacephei.com).
+
 ## Python installation from Pypi

 The easiest way to install vosk api is with pip. You do not have to compile anything. We currently support only Linux on x86_64 and Raspberry Pi. Other systems (windows, mac) will come soon.
@@ -31,6 +45,19 @@ Uprade python and pip if needed. Then install vosk on Linux with a simple comman
 pip3 install vosk
 ```

+## Websocket Server and GRPC server
+
+We also provide a websocket server and grpc server which can be used in telephony and other applications. With bigger models adapted for 8khz audio it provides more accuracy.
+
+The server is installed with docker and can run with a single command:
+
+```
+docker run -d -p 2700:2700 alphacep/kaldi-en:latest
+```
+
+For details see https://github.com/alphacep/vosk-server
+
+
 ## Compilation from source

 If you still want to build from scratch, you can compile Kaldi and Vosk yourself. The compilation is straightforward but might be a little confusing for newbie. In case you want to follow this, please watch the errors.
@@ -62,14 +89,30 @@ cd python
 python3 setup.py install
 ```

+#### Running the example code with python
+
+Run like this:
+
+```
+cd vosk-api/python/example
+wget https://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.zip
+unzip alphacep-model-android-en-us-0.3.zip
+mv alphacep-model-android-en-us-0.3 model-en
+python3 ./test_simple.py test.wav
+```
+
+To run with your audio file make sure it has proper format - PCM 16khz 16bit mono, otherwise decoding will not work.
+
+You can find other examples of using a microphone, decoding with a fixed small vocabulary or speaker identification setup in  [python/example subfolder](https://github.com/alphacep/vosk-api/tree/master/python/example)
+
 #### Java example API build

 Or Java

 ```
 cd java && KALDI_ROOT=<KALDI_ROOT> make
-wget https://github.com/alphacep/kaldi-android-demo/releases/download/2020-01/alphacep-model-android-en-us-0.3.tar.gz
-tar xf alphacep-model-android-en-us-0.3.tar.gz 
+wget https://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.zip
+unzip alphacep-model-android-en-us-0.3.zip
 mv alphacep-model-android-en-us-0.3 model
 make run
 ```
@@ -80,29 +123,15 @@ Or C#

 ```
 cd csharp && KALDI_ROOT=<KALDI_ROOT> make
-wget https://github.com/alphacep/kaldi-android-demo/releases/download/2020-01/alphacep-model-android-en-us-0.3.tar.gz
-tar xf alphacep-model-android-en-us-0.3.tar.gz 
+wget https://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.zip
+unzip alphacep-model-android-en-us-0.3.zip
 mv alphacep-model-android-en-us-0.3 model
 mono test.exe
 ```

-## Running the example code with python
+## Models for different languages

-Run like this:
-
-```
-cd vosk-api/python/example
-wget https://github.com/alphacep/kaldi-android-demo/releases/download/2020-01/alphacep-model-android-en-us-0.3.tar.gz
-tar xf alphacep-model-android-en-us-0.3.tar.gz 
-mv alphacep-model-android-en-us-0.3 model
-python3 ./test_simple.py test.wav
-```
-
-There are models for other languages (English, Chinese, Spanish, Portuguese, German, French, Russian) available too at https://github.com/alphacep/kaldi-android-demo/releases
-
-To run with your audio file make sure it has proper format - PCM 16khz 16bit mono, otherwise decoding will not work.
-
-You can find other examples of using a microphone, decoding with a fixed small vocabulary or speaker identification setup in  [python/example subfolder](https://github.com/alphacep/vosk-api/tree/master/python/example)
+For information about models see [the documentation on available models](https://github.com/alphacep/vosk-api/blob/master/doc/models.md).

 ## Contact Us

@@ -111,4 +140,3 @@ 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)
-   
@@ -0,0 +1,149 @@
+[![Build Status](https://travis-ci.com/alphacep/vosk-api.svg?branch=master)](https://travis-ci.com/alphacep/vosk-api)
+
+[EN](README.md)
+
+[中文](README.zh.md)
+
+Библитека для распознавания речи "Воск". Преимущества библиотеки:
+
+  1. Поддерживает 9 языков - русский, английский, немецкий, французский, португальский, испанский, китайский, турецкий, вьетнамский. В скором времени будут добавлены и другие
+  1. Работает без доступа к сети даже на мобильных устройствах - Raspberry Pi, Android, iOS
+  1. Устанавливается с помощью простой команды `pip3 install vosk` без дополнительных шагов
+  1. Модели для каждого языка занимают всего 50Мб, но есть и гораздо более точные большие модели для более точного распознавания
+  1. Сделана для потоковой обработки звука, что позволяет реализовать мгновенную реакцию на команды
+  1. Поддерживает несколько популярных языков программирования - Java, C#, Javascript
+  1. Позволяет быстро настраивать словарь распознавания для улучшения точности распознавания
+  1. Позволяет идентифицировать говорящего
+
+## Сборка для Android
+
+```
+cd android
+gradle build
+```
+
+Сборка включает платформы x86, armv7, arm64
+
+Для примера приложения, созданного с помощью библиотеки "Воск" смотрите [демо проект](https://github.com/alphacep/kaldi-android-demo).
+
+## Сборка для iOS
+
+Доступна позапросу. Напишите нам [contact@alphacephei.com](mailto:contact@alphacephei.com).
+
+## Установка для работы Python из Pypi
+
+Проще всего установить "Воск" с помощью pip. Собирать ничего не нужно. Мы поддерживаем платформы Linux, RPi и Windows. Сборка для OSX будет скоро доступна.
+
+Для начала убедитесь, что используются достаточно новые версии pip и Python:
+
+  * Python версия >= 3.5
+  * pip версия >= 19.0
+
+Обновите Python и Pip если нужно, а затем установите "Воск" такой командой:
+
+```
+pip3 install vosk
+```
+
+Для использования "Воск" смотрите примеры ниже.
+
+## Сервер для протоколов Websocket и GRPC
+
+We also provide a websocket server and grpc server which can be used in telephony and other applications. With bigger models adapted for 8khz audio it provides more accuracy.
+
+The server is installed with docker and can run with a single command:
+
+```
+docker run -d -p 2700:2700 alphacep/kaldi-en:latest
+```
+
+Смотрите проект https://github.com/alphacep/vosk-server
+
+
+## Сборка из исходников
+
+Если нужно собрать проект из исходного кода, необходимо будет собрать 
+Kaldi самостоятельно. Сборка досаточно простая и прямолинейная, но может
+быть непривычной для начинающих. Обращайте внимания на сообщения об ошибках
+в процессе сборки.
+
+#### Сборка Kaldi для модулей на Python, Java, C#
+
+```
+git clone https://github.com/kaldi-asr/kaldi
+cd kaldi/tools
+make
+```
+
+установите все рекомандуемые пакеты и повторите `make` если потребуется.
+
+```
+extras/install_openblas.sh
+cd ../src
+./configure --mathlib=OPENBLAS --shared --use-cuda=no
+make -j 10
+```
+
+#### Сборка модуля на Python
+
+После Kaldi можно собрать модуль Python
+
+```
+export KALDI_ROOT=<KALDI_ROOT>
+cd python
+python3 setup.py install
+```
+
+#### Запуск примера для Python
+
+Выполните следующие команды:
+
+```
+cd vosk-api/python/example
+wget https://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.zip
+unzip alphacep-model-android-en-us-0.3.zip
+mv alphacep-model-android-en-us-0.3 model-en
+python3 ./test_simple.py test.wav
+```
+
+Для того, чтобы распознавать другой файл, переведите его в нужный формат - PCM 16кГц 16бит 1канал. Это можно сделать с помощью ffmpeg.
+
+Другие примеры, в том числе использования микрофона, распознавание с небольшим словарём и распознавание говорящего можно найти в [каталоге python/example](https://github.com/alphacep/vosk-api/tree/master/python/example)
+
+#### Сборка для Java
+
+Перейдите в каталог Java и запустите сборку
+
+```
+cd java && KALDI_ROOT=<KALDI_ROOT> make
+wget https://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.zip
+unzip alphacep-model-android-en-us-0.3.zip
+mv alphacep-model-android-en-us-0.3 model
+make run
+```
+
+#### Сборка для C#
+
+Для сборки в среде Mono.
+
+```
+cd csharp && KALDI_ROOT=<KALDI_ROOT> make
+wget https://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.zip
+unzip alphacep-model-android-en-us-0.3.zip
+mv alphacep-model-android-en-us-0.3 model
+mono test.exe
+```
+
+.NET тоже должен работать, хотя мы не пробовали.
+
+## Модели для разных языков
+
+По информации о моделях смотрите соответствующую [страницу документации](https://github.com/alphacep/vosk-api/blob/master/doc/models.md).
+
+## Contact Us
+
+Если возникли вопросы, свяжитесь с нами:
+
+   * Создайте проблему тут на github
+   * Напишите нам по почте [contact@alphacephei.com](mailto:contact@alphacephei.com)
+   * Заходите в нашу группу в Телеграмме [@speech_recognition_ru](https://t.me/speech_recognition_ru)
@@ -0,0 +1,12 @@
+[![Build Status](https://travis-ci.com/alphacep/vosk-api.svg?branch=master)](https://travis-ci.com/alphacep/vosk-api)
+
+Vosk是言语识别工具包。Vosk最好的事情是：
+
+  1. 支持九种语言 - 中文, 英语，德语，法语，西班牙语，葡萄牙语，俄语，土耳其语，越南语
+  1. 移动设备上脱机工作-Raspberry Pi，Android，iOS
+  1. 使用简单的 pip3 install vosk 安装
+  1. 每种语言的手提式模型只有是50Mb, 但还有更大的服务器模型可用
+  1. 提供流媒体API，以提供最佳用户体验（与流行的语音识别python包不同）
+  1. 还有用于不同编程语言的包装器-java / csharp / javascript等
+  1. 可以快速重新配置词汇以实现最佳准确性
+  1. 支持说话人识别
@@ -23,6 +23,8 @@ set(API_SOURCES
 "${PROJECT_SOURCE_DIR}/../src/model.h"
 "${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")
@@ -43,6 +45,7 @@ target_link_libraries( kaldi_jni
             ${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
             ${KALDI_ROOT}/src/cudamatrix/kaldi-cudamatrix.a
@@ -125,6 +125,6 @@ CXX=$CXX CXXFLAGS="$ARCHFLAGS -O3 -DFST_NO_DYNAMIC_LINKING" ./configure --use-cu
    --fst-root=${WORKDIR}/local --fst-version=${OPENFST_VERSION}

 make -j 8 depend
-make -j 8 online2
+make -j 8 online2 lm

 done
@@ -1,3 +1,3 @@
 <?xml version="1.0" encoding="utf-8"?>
-<manifest xmlns:android="http://schemas.android.com/apk/res/android" package="edu.cmu.pocketsphinx">
+<manifest xmlns:android="http://schemas.android.com/apk/res/android" package="org.kaldi">
 </manifest>
@@ -241,10 +241,6 @@ public class Assets {
            if (!items.get(path).equals(externalItems.get(path))
                    || !(new File(externalDir, path).exists()))
                newItems.add(path);
-            else
-                Log.i(TAG,
-                        String.format("Skipping asset %s: checksums are equal", path));
-
        }

        unusedItems.addAll(externalItems.keySet());
@@ -252,13 +248,11 @@ public class Assets {

        for (String path : newItems) {
            File file = copy(path);
-            Log.i(TAG, String.format("Copying asset %s to %s", path, file));
        }

        for (String path : unusedItems) {
            File file = new File(externalDir, path);
            file.delete();
-            Log.i(TAG, String.format("Removing asset %s", file));
        }

        updateItemList(items);
@@ -31,7 +31,7 @@ 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
- * Pocketsphinx engine. Recognition events are passed to a client using
+ * VOSK engine. Recognition events are passed to a client using
 * {@link RecognitionListener}
 * 
 */
@@ -148,8 +148,7 @@ public class SpeechRecognizer {
    public boolean stop() {
        boolean result = stopRecognizerThread();
        if (result) {
-            Log.i(TAG, "Stop recognition");
-            mainHandler.post(new ResultEvent(recognizer.FinalResult(), true));
+            mainHandler.post(new ResultEvent(recognizer.Result(), true));
        }
        return result;
    }
@@ -162,10 +161,7 @@ public class SpeechRecognizer {
     */
    public boolean cancel() {
        boolean result = stopRecognizerThread();
-        if (result) {
-            Log.i(TAG, "Cancel recognition");
-        }
-
+        recognizer.Result(); // Reset recognizer state
        return result;
    }
    
@@ -206,8 +202,6 @@ public class SpeechRecognizer {
                return;
            }

-            Log.d(TAG, "Starting decoding");
-
            short[] buffer = new short[bufferSize];

            while (!interrupted()
@@ -11,6 +11,7 @@ KALDI_LIBS = \
             ${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 \
@@ -21,7 +22,7 @@ KALDI_LIBS = \
             ${KALDI_ROOT}/tools/openfst/lib/libfst.a \
             ${KALDI_ROOT}/tools/openfst/lib/libfstngram.a \
             ${KALDI_ROOT}/tools/OpenBLAS/libopenblas.a \
-             -lgfortran
+             -lgfortran -lstdc++

 all: test.exe

@@ -29,20 +30,25 @@ test.exe: libkaldiwrap.so test.cs
 	mcs test.cs gen/*.cs

 VOSK_SOURCES = \
-	vosk_wrap.cc \
+	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/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.cc: ../src/vosk.i
-	swig -csharp -dllimport "libkaldiwrap.so" \
-	    -namespace "Kaldi" -c++ -outdir gen -o vosk_wrap.cc ../src/vosk.i
+vosk_wrap.c: ../src/vosk.i
+	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.cc *.o gen/*.cs test.exe
+	$(RM) *.so vosk_wrap.c *.o gen/*.cs test.exe
@@ -7,6 +7,7 @@ public class Test
   public static void Main()
   {

+        Vosk.SetLogLevel(0);
        Model model = new Model("model");
        KaldiRecognizer rec = new KaldiRecognizer(model, 16000.0f);

@@ -0,0 +1,21 @@
+## Accuracy issues
+
+Accuracy of modern systems is still unstable, that means sometimes you can have a very good accuracy and sometimes it could be bad.
+It is hard to make a system that will work good. And there could be many reasons for that:
+
+  * Audio has very bad quality
+  * Vocabulary of the system doesn't match (yes, we still use fixed vocabulary)
+  * Audio conditions like accent were not really the ones that were used in training
+  * Some unpredictable audio issues like frame drop or frame coding bugs
+  * Software bugs
+
+It is hard to guess what is going on under the hood without getting your hands dirty. For that reason in case of any accuracy
+issues you must provide the following for analysis:
+
+  * Who are you, where are you from and why are you doing that. We don't like dealing with anonymous
+  * The complete and exact description of the system you want to build - what is it going to do, what do you want to build
+  * The precise description of hardware you are trying to run the system on
+  * The detailed list of software versions you are using
+  * Audio samples to demonstrate the problem together with the reference transcription for those samples
+
+Remember, the more information you provide the faster you get a solution.
@@ -0,0 +1,81 @@
+# Models
+
+This is the list of models compatible with Vosk-API.
+
+To add a new model here create an issue on Github.
+
+| Model                                                                                                     | Size  | Accuracy   | Notes                                                                                        |
+|-------------------------------------------------------------------------------------------------------------------------|-------|------------|----------------------------------------------------------------------------------------------|
+| **English**                                                                                               |       |            |  |
+| [vosk-model-en-us-aspire-0.2](http://alphacephei.com/kaldi/models/vosk-model-en-us-aspire-0.2.zip)        |  1.4G |   TBD      | Trained on Fisher + more or less recent LM. Should be pretty good for generic US English transcription |
+| [vosk-model-small-en-us-0.3](http://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.zip)          |  36M  |   TBD      | Lightweight wideband model for Android and RPi                                               |
+| **Chinese**                                                                                                   |       |            |  |
+| [vosk-model-cn-0.1.zip](http://alphacephei.com/kaldi/models/vosk-model-cn-0.1.zip)                        |  195M |   TBD      | Big narrowband Chinese model for server processing                                           |
+| [vosk-model-small-cn-0.3](http://alphacephei.com/kaldi/models/vosk-model-small-cn-0.3.zip)                |  32M  |   TBD      | Lightweight wideband model for Android and RPi                                               |
+| **Russian**                                                                                                   |       |            |  |
+| [vosk-model-ru-0.10.zip](http://alphacephei.com/kaldi/models/vosk-model-ru-0.10.zip)                      |  2.5G |   TBD      | Big narrowband Russian model for server processing                                           |
+| [vosk-model-small-ru-0.4](http://alphacephei.com/kaldi/models/vosk-model-small-ru-0.4.zip)                |  39M  |   TBD      | Lightweight wideband model for Android and RPi                                               |
+| **French**                                                                                                |       |            |  |
+| [vosk-model-small-fr-pguyot-0.3](http://alphacephei.com/kaldi/models/vosk-model-small-fr-pguyot-0.3.zip)                |  39M  |   TBD      | Lightweight wideband model for Android and RPi trained by [Paul Guyot](https://github.com/pguyot/zamia-speech/releases) |
+| **German**                                                                                                |       |            |  |
+| [tuda-de](http://ltdata1.informatik.uni-hamburg.de/kaldi_tuda_de/de_400k_nnet3chain_tdnn1f_2048_sp_bi.tar.bz2)    |  566M |   TBD      | Wideband server model from [tuda-de](https://github.com/uhh-lt/vosk-model-tuda-de)            |
+| [vosk-model-small-de-zamia-0.3](http://alphacephei.com/kaldi/models/vosk-model-small-de-zamia-0.3.zip)            |  49M  |   TBD      | Lightweight wideband model for Android and RPi                                           |
+| **Spanish**                                                                                                |       |            |  |
+| [vosk-model-small-es-0.3](http://alphacephei.com/kaldi/models/vosk-model-small-es-0.3.zip)                |  33M  |   TBD      | Lightweight wideband model for Android and RPi                                               |
+| **Portuguese**                                                                                                |       |            |  |
+| [vosk-model-small-pt-0.3](http://alphacephei.com/kaldi/models/vosk-model-small-pt-0.3.zip)                |  31M  |   TBD      | Lightweight wideband model for Android and RPi                                               |
+| **Greek**                                                                                                |       |            |  |
+| [vosk-model-el-gr-0.6.zip](http://alphacephei.com/kaldi/models/vosk-model-el-gr-0.6.zip)                  |  1.1G |   TBD      | Big narrowband Greek model for server processing, not extremely accurate though        |
+| **Turkish**                                                                                                |       |            |  |
+| [vosk-model-small-tr-0.3](http://alphacephei.com/kaldi/models/vosk-model-small-tr-0.3.zip)                |  35M  |   TBD      | Lightweight wideband model for Android and RPi                                               |
+| **Vietnamese**                                                                                                |       |            |  |
+| [vosk-model-small-vn-0.3](http://alphacephei.com/kaldi/models/vosk-model-small-vn-0.3.zip)                |  32M  |   TBD      | Lightweight wideband model for Android and RPi                                               |
+| **Speaker identification model**                                                                          |       |            |                                                                                              |
+| [vosk-model-spk-0.3](http://alphacephei.com/kaldi/models/vosk-model-spk-0.3.zip)                          |  13M  |   TBD      | Model for speaker identification, should work for all languages                              |
+
+
+## Other models
+
+Other places where you can check for models which might be compatible:
+
+  * http://kaldi-asr.org/models.html - variety of models from Kaldi - librispeech, aspire, chinese models
+  * https://github.com/daanzu/kaldi-active-grammar/blob/master/docs/models.md - Big dictation models
+  * http://zamia-speech.org/asr/ - German and English model from Zamia
+  * https://github.com/pguyot/zamia-speech/releases - French models for Zamia
+  * https://github.com/opensource-spraakherkenning-nl/Kaldi_NL - Dutch model
+  * https://montreal-forced-aligner.readthedocs.io/en/latest/pretrained_models.html (GMM models, not compatible but might be still useful)
+  * https://github.com/goodatlas/zeroth - Korean Kaldi (just a recipe and data to train)
+  * https://github.com/undertheseanlp/automatic_speech_recognition - Vietnamese Kaldi project
+
+## Training your own model
+
+You can train your model with Kaldi toolkit. The training is pretty standard - you need tdnn nnet3 model with ivectors. You can
+check mini_librispeech recipe for details. Some notes on training:
+
+  * For smaller mobile models watch number of parameters
+  * Train the model without pitch. It might be helpful for small amount of data, but for large database it doesn't give the advantage
+but complicates the processing and increases response time.
+  * Train ivector of dim 30 instead of standard 100 to save memory of mobile models.
+
+## Model structure
+
+Once you trained the model arrange the files according to the following layout (see en-us-aspire for details):
+
+  * `am/final.mdl` - acoustic model
+  * `conf/mfcc.conf` - mfcc config file. Make sure you take mfcc_hires.conf version if you are using hires model (most external ones)
+  * `conf/model.conf` - provide default decoding beams and silence phones. you have to create this file yourself, it is not present in kaldi model
+  * `ivector/final.dubm` - take ivector files from ivector extractor
+  * `ivector/final.ie`
+  * `ivector/final.mat`
+  * `ivector/splice.conf`
+  * `ivector/global_cmvn.stats`
+  * `ivector/online_cmvn.conf`
+  * `graph/phones/word_boundary.int` - from the graph
+  * `graph/HCLG.fst` - this is the decoding graph, if you are not using lookahead
+  * `graph/HCLr.fst` - use Gr.fst and HCLr.fst instead of one big HCLG.fst if you want to run rescoring
+  * `graph/Gr.fst`
+  * `graph/phones.txt` - from the graph
+  * `graph/words.txt` - from the graph
+  * `rescore/G.carpa` - carpa rescoring is optional but helpful in big models. Usually located inside data/lang_test_rescore
+  * `rescore/G.fst` - also optional if you want to use rescoring
+
@@ -0,0 +1,3 @@
+This is a baseline for the vosk-api iOS demo. It requires a build of a
+Vosk-API library, mail contact@alphacephei.com for the details.
+
@@ -0,0 +1,477 @@
+// !$*UTF8*$!
+{
+	archiveVersion = 1;
+	classes = {
+	};
+	objectVersion = 46;
+	objects = {
+
+/* Begin PBXBuildFile section */
+		92375222240C550B00DD6076 /* AppDelegate.swift in Sources */ = {isa = PBXBuildFile; fileRef = 92375221240C550B00DD6076 /* AppDelegate.swift */; };
+		92375224240C550B00DD6076 /* ViewController.swift in Sources */ = {isa = PBXBuildFile; fileRef = 92375223240C550B00DD6076 /* ViewController.swift */; };
+		92375227240C550B00DD6076 /* Main.storyboard in Resources */ = {isa = PBXBuildFile; fileRef = 92375225240C550B00DD6076 /* Main.storyboard */; };
+		92375229240C550B00DD6076 /* Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = 92375228240C550B00DD6076 /* Assets.xcassets */; };
+		9237522C240C550B00DD6076 /* LaunchScreen.storyboard in Resources */ = {isa = PBXBuildFile; fileRef = 9237522A240C550B00DD6076 /* LaunchScreen.storyboard */; };
+		92375234240C558900DD6076 /* Vosk.swift in Sources */ = {isa = PBXBuildFile; fileRef = 92375233240C558900DD6076 /* Vosk.swift */; };
+		9237523C240C642000DD6076 /* libkaldiwrap.a in Frameworks */ = {isa = PBXBuildFile; fileRef = 9237523A240C642000DD6076 /* libkaldiwrap.a */; };
+		92375244240C6DAF00DD6076 /* Accelerate.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 92375243240C6DAF00DD6076 /* Accelerate.framework */; };
+		92375246240C6DC900DD6076 /* libstdc++.tbd in Frameworks */ = {isa = PBXBuildFile; fileRef = 92375245240C6DC900DD6076 /* libstdc++.tbd */; };
+		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 */; };
+/* 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>"; };
+		92375223240C550B00DD6076 /* ViewController.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = ViewController.swift; sourceTree = "<group>"; };
+		92375226240C550B00DD6076 /* Base */ = {isa = PBXFileReference; lastKnownFileType = file.storyboard; name = Base; path = Base.lproj/Main.storyboard; sourceTree = "<group>"; };
+		92375228240C550B00DD6076 /* Assets.xcassets */ = {isa = PBXFileReference; lastKnownFileType = folder.assetcatalog; path = Assets.xcassets; sourceTree = "<group>"; };
+		9237522B240C550B00DD6076 /* Base */ = {isa = PBXFileReference; lastKnownFileType = file.storyboard; name = Base; path = Base.lproj/LaunchScreen.storyboard; sourceTree = "<group>"; };
+		9237522D240C550B00DD6076 /* Info.plist */ = {isa = PBXFileReference; lastKnownFileType = text.plist.xml; path = Info.plist; sourceTree = "<group>"; };
+		92375233240C558900DD6076 /* Vosk.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; path = Vosk.swift; sourceTree = "<group>"; };
+		9237523A240C642000DD6076 /* libkaldiwrap.a */ = {isa = PBXFileReference; lastKnownFileType = archive.ar; path = libkaldiwrap.a; sourceTree = "<group>"; };
+		92375243240C6DAF00DD6076 /* Accelerate.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = Accelerate.framework; path = System/Library/Frameworks/Accelerate.framework; sourceTree = SDKROOT; };
+		92375245240C6DC900DD6076 /* libstdc++.tbd */ = {isa = PBXFileReference; lastKnownFileType = "sourcecode.text-based-dylib-definition"; name = "libstdc++.tbd"; path = "usr/lib/libstdc++.tbd"; sourceTree = SDKROOT; };
+		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>"; };
+		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>"; };
+/* End PBXFileReference section */
+
+/* Begin PBXFrameworksBuildPhase section */
+		9237521B240C550B00DD6076 /* Frameworks */ = {
+			isa = PBXFrameworksBuildPhase;
+			buildActionMask = 2147483647;
+			files = (
+				92375246240C6DC900DD6076 /* libstdc++.tbd in Frameworks */,
+				92375244240C6DAF00DD6076 /* Accelerate.framework in Frameworks */,
+				9237523C240C642000DD6076 /* libkaldiwrap.a in Frameworks */,
+			);
+			runOnlyForDeploymentPostprocessing = 0;
+		};
+/* End PBXFrameworksBuildPhase section */
+
+/* Begin PBXGroup section */
+		92375215240C550A00DD6076 = {
+			isa = PBXGroup;
+			children = (
+				92375239240C642000DD6076 /* Vosk */,
+				92375220240C550B00DD6076 /* VoskApiTest */,
+				9237521F240C550B00DD6076 /* Products */,
+				92375242240C6DAF00DD6076 /* Frameworks */,
+			);
+			sourceTree = "<group>";
+		};
+		9237521F240C550B00DD6076 /* Products */ = {
+			isa = PBXGroup;
+			children = (
+				9237521E240C550B00DD6076 /* VoskApiTest.app */,
+			);
+			name = Products;
+			sourceTree = "<group>";
+		};
+		92375220240C550B00DD6076 /* VoskApiTest */ = {
+			isa = PBXGroup;
+			children = (
+				92AA22AE244CDD5200DA464B /* bridging.h */,
+				92375221240C550B00DD6076 /* AppDelegate.swift */,
+				92375223240C550B00DD6076 /* ViewController.swift */,
+				92375225240C550B00DD6076 /* Main.storyboard */,
+				92375228240C550B00DD6076 /* Assets.xcassets */,
+				9237522A240C550B00DD6076 /* LaunchScreen.storyboard */,
+				9237522D240C550B00DD6076 /* Info.plist */,
+				92375233240C558900DD6076 /* Vosk.swift */,
+			);
+			path = VoskApiTest;
+			sourceTree = "<group>";
+		};
+		92375239240C642000DD6076 /* Vosk */ = {
+			isa = PBXGroup;
+			children = (
+				92375247240C6E3D00DD6076 /* model-android */,
+				92375256240C6E3D00DD6076 /* 10001-90210-01803.wav */,
+				92AA22AD244CDD1200DA464B /* vosk_api.h */,
+				9237523A240C642000DD6076 /* libkaldiwrap.a */,
+			);
+			name = Vosk;
+			path = VoskApiTest/Vosk;
+			sourceTree = "<group>";
+		};
+		92375242240C6DAF00DD6076 /* Frameworks */ = {
+			isa = PBXGroup;
+			children = (
+				92375245240C6DC900DD6076 /* libstdc++.tbd */,
+				92375243240C6DAF00DD6076 /* Accelerate.framework */,
+			);
+			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 */
+		9237521D240C550B00DD6076 /* VoskApiTest */ = {
+			isa = PBXNativeTarget;
+			buildConfigurationList = 92375230240C550B00DD6076 /* Build configuration list for PBXNativeTarget "VoskApiTest" */;
+			buildPhases = (
+				9237521A240C550B00DD6076 /* Sources */,
+				9237521B240C550B00DD6076 /* Frameworks */,
+				9237521C240C550B00DD6076 /* Resources */,
+				92375265240C6ECF00DD6076 /* CopyFiles */,
+				9237526D240C6F0400DD6076 /* CopyFiles */,
+			);
+			buildRules = (
+			);
+			dependencies = (
+			);
+			name = VoskApiTest;
+			productName = VoskApiTest;
+			productReference = 9237521E240C550B00DD6076 /* VoskApiTest.app */;
+			productType = "com.apple.product-type.application";
+		};
+/* End PBXNativeTarget section */
+
+/* Begin PBXProject section */
+		92375216240C550A00DD6076 /* Project object */ = {
+			isa = PBXProject;
+			attributes = {
+				LastSwiftUpdateCheck = 0830;
+				LastUpgradeCheck = 0920;
+				ORGANIZATIONNAME = "Alpha Cephei";
+				TargetAttributes = {
+					9237521D240C550B00DD6076 = {
+						CreatedOnToolsVersion = 8.3.2;
+						LastSwiftMigration = 0920;
+						ProvisioningStyle = Automatic;
+					};
+				};
+			};
+			buildConfigurationList = 92375219240C550A00DD6076 /* Build configuration list for PBXProject "VoskApiTest" */;
+			compatibilityVersion = "Xcode 3.2";
+			developmentRegion = English;
+			hasScannedForEncodings = 0;
+			knownRegions = (
+				en,
+				Base,
+			);
+			mainGroup = 92375215240C550A00DD6076;
+			productRefGroup = 9237521F240C550B00DD6076 /* Products */;
+			projectDirPath = "";
+			projectRoot = "";
+			targets = (
+				9237521D240C550B00DD6076 /* VoskApiTest */,
+			);
+		};
+/* End PBXProject section */
+
+/* Begin PBXResourcesBuildPhase section */
+		9237521C240C550B00DD6076 /* Resources */ = {
+			isa = PBXResourcesBuildPhase;
+			buildActionMask = 2147483647;
+			files = (
+				92375274240C6F1E00DD6076 /* 10001-90210-01803.wav in Resources */,
+				9237522C240C550B00DD6076 /* LaunchScreen.storyboard in Resources */,
+				92375229240C550B00DD6076 /* Assets.xcassets in Resources */,
+				92375227240C550B00DD6076 /* Main.storyboard in Resources */,
+			);
+			runOnlyForDeploymentPostprocessing = 0;
+		};
+/* End PBXResourcesBuildPhase section */
+
+/* Begin PBXSourcesBuildPhase section */
+		9237521A240C550B00DD6076 /* Sources */ = {
+			isa = PBXSourcesBuildPhase;
+			buildActionMask = 2147483647;
+			files = (
+				92375224240C550B00DD6076 /* ViewController.swift in Sources */,
+				92375222240C550B00DD6076 /* AppDelegate.swift in Sources */,
+				92375234240C558900DD6076 /* Vosk.swift in Sources */,
+			);
+			runOnlyForDeploymentPostprocessing = 0;
+		};
+/* End PBXSourcesBuildPhase section */
+
+/* Begin PBXVariantGroup section */
+		92375225240C550B00DD6076 /* Main.storyboard */ = {
+			isa = PBXVariantGroup;
+			children = (
+				92375226240C550B00DD6076 /* Base */,
+			);
+			name = Main.storyboard;
+			sourceTree = "<group>";
+		};
+		9237522A240C550B00DD6076 /* LaunchScreen.storyboard */ = {
+			isa = PBXVariantGroup;
+			children = (
+				9237522B240C550B00DD6076 /* Base */,
+			);
+			name = LaunchScreen.storyboard;
+			sourceTree = "<group>";
+		};
+/* End PBXVariantGroup section */
+
+/* Begin XCBuildConfiguration section */
+		9237522E240C550B00DD6076 /* Debug */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				ALWAYS_SEARCH_USER_PATHS = NO;
+				CLANG_ANALYZER_NONNULL = YES;
+				CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
+				CLANG_CXX_LANGUAGE_STANDARD = "gnu++0x";
+				CLANG_CXX_LIBRARY = "libc++";
+				CLANG_ENABLE_MODULES = YES;
+				CLANG_ENABLE_OBJC_ARC = YES;
+				CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
+				CLANG_WARN_BOOL_CONVERSION = YES;
+				CLANG_WARN_COMMA = YES;
+				CLANG_WARN_CONSTANT_CONVERSION = YES;
+				CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
+				CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
+				CLANG_WARN_EMPTY_BODY = YES;
+				CLANG_WARN_ENUM_CONVERSION = YES;
+				CLANG_WARN_INFINITE_RECURSION = YES;
+				CLANG_WARN_INT_CONVERSION = YES;
+				CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
+				CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
+				CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
+				CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
+				CLANG_WARN_STRICT_PROTOTYPES = YES;
+				CLANG_WARN_SUSPICIOUS_MOVE = YES;
+				CLANG_WARN_UNREACHABLE_CODE = YES;
+				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
+				"CODE_SIGN_IDENTITY[sdk=iphoneos*]" = "iPhone Developer";
+				COPY_PHASE_STRIP = NO;
+				DEBUG_INFORMATION_FORMAT = dwarf;
+				ENABLE_STRICT_OBJC_MSGSEND = YES;
+				ENABLE_TESTABILITY = YES;
+				GCC_C_LANGUAGE_STANDARD = gnu99;
+				GCC_DYNAMIC_NO_PIC = NO;
+				GCC_NO_COMMON_BLOCKS = YES;
+				GCC_OPTIMIZATION_LEVEL = 0;
+				GCC_PREPROCESSOR_DEFINITIONS = (
+					"DEBUG=1",
+					"$(inherited)",
+				);
+				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
+				GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
+				GCC_WARN_UNDECLARED_SELECTOR = YES;
+				GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
+				GCC_WARN_UNUSED_FUNCTION = YES;
+				GCC_WARN_UNUSED_VARIABLE = YES;
+				IPHONEOS_DEPLOYMENT_TARGET = 10.3;
+				MTL_ENABLE_DEBUG_INFO = YES;
+				ONLY_ACTIVE_ARCH = YES;
+				SDKROOT = iphoneos;
+				SWIFT_ACTIVE_COMPILATION_CONDITIONS = DEBUG;
+				SWIFT_OBJC_BRIDGING_HEADER = bridging.h;
+				SWIFT_OPTIMIZATION_LEVEL = "-Onone";
+				SWIFT_SWIFT3_OBJC_INFERENCE = Default;
+				TARGETED_DEVICE_FAMILY = "1,2";
+			};
+			name = Debug;
+		};
+		9237522F240C550B00DD6076 /* Release */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				ALWAYS_SEARCH_USER_PATHS = NO;
+				CLANG_ANALYZER_NONNULL = YES;
+				CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
+				CLANG_CXX_LANGUAGE_STANDARD = "gnu++0x";
+				CLANG_CXX_LIBRARY = "libc++";
+				CLANG_ENABLE_MODULES = YES;
+				CLANG_ENABLE_OBJC_ARC = YES;
+				CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
+				CLANG_WARN_BOOL_CONVERSION = YES;
+				CLANG_WARN_COMMA = YES;
+				CLANG_WARN_CONSTANT_CONVERSION = YES;
+				CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
+				CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
+				CLANG_WARN_EMPTY_BODY = YES;
+				CLANG_WARN_ENUM_CONVERSION = YES;
+				CLANG_WARN_INFINITE_RECURSION = YES;
+				CLANG_WARN_INT_CONVERSION = YES;
+				CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
+				CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
+				CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
+				CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
+				CLANG_WARN_STRICT_PROTOTYPES = YES;
+				CLANG_WARN_SUSPICIOUS_MOVE = YES;
+				CLANG_WARN_UNREACHABLE_CODE = YES;
+				CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
+				"CODE_SIGN_IDENTITY[sdk=iphoneos*]" = "iPhone Developer";
+				COPY_PHASE_STRIP = NO;
+				DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
+				ENABLE_NS_ASSERTIONS = NO;
+				ENABLE_STRICT_OBJC_MSGSEND = YES;
+				GCC_C_LANGUAGE_STANDARD = gnu99;
+				GCC_NO_COMMON_BLOCKS = YES;
+				GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
+				GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
+				GCC_WARN_UNDECLARED_SELECTOR = YES;
+				GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
+				GCC_WARN_UNUSED_FUNCTION = YES;
+				GCC_WARN_UNUSED_VARIABLE = YES;
+				IPHONEOS_DEPLOYMENT_TARGET = 10.3;
+				MTL_ENABLE_DEBUG_INFO = NO;
+				SDKROOT = iphoneos;
+				SWIFT_OBJC_BRIDGING_HEADER = bridging.h;
+				SWIFT_OPTIMIZATION_LEVEL = "-Owholemodule";
+				SWIFT_SWIFT3_OBJC_INFERENCE = Default;
+				TARGETED_DEVICE_FAMILY = "1,2";
+				VALIDATE_PRODUCT = YES;
+			};
+			name = Release;
+		};
+		92375231240C550B00DD6076 /* Debug */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
+				CLANG_ENABLE_MODULES = YES;
+				ENABLE_BITCODE = NO;
+				INFOPLIST_FILE = VoskApiTest/Info.plist;
+				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
+				LIBRARY_SEARCH_PATHS = (
+					"$(inherited)",
+					"$(PROJECT_DIR)/VoskApiTest/Vosk",
+				);
+				PRODUCT_BUNDLE_IDENTIFIER = com.alphacephei.VoskApiTest;
+				PRODUCT_NAME = "$(TARGET_NAME)";
+				SWIFT_INSTALL_OBJC_HEADER = YES;
+				SWIFT_OBJC_BRIDGING_HEADER = VoskApiTest/bridging.h;
+				SWIFT_OPTIMIZATION_LEVEL = "-Onone";
+				SWIFT_SWIFT3_OBJC_INFERENCE = Default;
+				SWIFT_VERSION = 4.0;
+			};
+			name = Debug;
+		};
+		92375232240C550B00DD6076 /* Release */ = {
+			isa = XCBuildConfiguration;
+			buildSettings = {
+				ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
+				CLANG_ENABLE_MODULES = YES;
+				ENABLE_BITCODE = NO;
+				INFOPLIST_FILE = VoskApiTest/Info.plist;
+				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
+				LIBRARY_SEARCH_PATHS = (
+					"$(inherited)",
+					"$(PROJECT_DIR)/VoskApiTest/Vosk",
+				);
+				PRODUCT_BUNDLE_IDENTIFIER = com.alphacephei.VoskApiTest;
+				PRODUCT_NAME = "$(TARGET_NAME)";
+				SWIFT_INSTALL_OBJC_HEADER = YES;
+				SWIFT_OBJC_BRIDGING_HEADER = VoskApiTest/bridging.h;
+				SWIFT_SWIFT3_OBJC_INFERENCE = Default;
+				SWIFT_VERSION = 4.0;
+			};
+			name = Release;
+		};
+/* End XCBuildConfiguration section */
+
+/* Begin XCConfigurationList section */
+		92375219240C550A00DD6076 /* Build configuration list for PBXProject "VoskApiTest" */ = {
+			isa = XCConfigurationList;
+			buildConfigurations = (
+				9237522E240C550B00DD6076 /* Debug */,
+				9237522F240C550B00DD6076 /* Release */,
+			);
+			defaultConfigurationIsVisible = 0;
+			defaultConfigurationName = Release;
+		};
+		92375230240C550B00DD6076 /* Build configuration list for PBXNativeTarget "VoskApiTest" */ = {
+			isa = XCConfigurationList;
+			buildConfigurations = (
+				92375231240C550B00DD6076 /* Debug */,
+				92375232240C550B00DD6076 /* Release */,
+			);
+			defaultConfigurationIsVisible = 0;
+			defaultConfigurationName = Release;
+		};
+/* End XCConfigurationList section */
+	};
+	rootObject = 92375216240C550A00DD6076 /* Project object */;
+}
@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<Workspace
+   version = "1.0">
+   <FileRef
+      location = "self:VoskApiTest.xcodeproj">
+   </FileRef>
+</Workspace>
@@ -0,0 +1,46 @@
+//
+//  AppDelegate.swift
+//  VoskApiTest
+//
+//  Created by Niсkolay Shmyrev on 01.03.20.
+//  Copyright © 2020 Alpha Cephei. All rights reserved.
+//
+
+import UIKit
+
+@UIApplicationMain
+class AppDelegate: UIResponder, UIApplicationDelegate {
+
+    var window: UIWindow?
+
+
+    func application(_ application: UIApplication, didFinishLaunchingWithOptions launchOptions: [UIApplicationLaunchOptionsKey: Any]?) -> Bool {
+        // Override point for customization after application launch.
+        return true
+    }
+
+    func applicationWillResignActive(_ application: UIApplication) {
+        // Sent when the application is about to move from active to inactive state. This can occur for certain types of temporary interruptions (such as an incoming phone call or SMS message) or when the user quits the application and it begins the transition to the background state.
+        // Use this method to pause ongoing tasks, disable timers, and invalidate graphics rendering callbacks. Games should use this method to pause the game.
+    }
+
+    func applicationDidEnterBackground(_ application: UIApplication) {
+        // Use this method to release shared resources, save user data, invalidate timers, and store enough application state information to restore your application to its current state in case it is terminated later.
+        // If your application supports background execution, this method is called instead of applicationWillTerminate: when the user quits.
+    }
+
+    func applicationWillEnterForeground(_ application: UIApplication) {
+        // Called as part of the transition from the background to the active state; here you can undo many of the changes made on entering the background.
+    }
+
+    func applicationDidBecomeActive(_ application: UIApplication) {
+        // Restart any tasks that were paused (or not yet started) while the application was inactive. If the application was previously in the background, optionally refresh the user interface.
+    }
+
+    func applicationWillTerminate(_ application: UIApplication) {
+        // Called when the application is about to terminate. Save data if appropriate. See also applicationDidEnterBackground:.
+    }
+
+
+}
+
@@ -0,0 +1,98 @@
+{
+  "images" : [
+    {
+      "idiom" : "iphone",
+      "size" : "20x20",
+      "scale" : "2x"
+    },
+    {
+      "idiom" : "iphone",
+      "size" : "20x20",
+      "scale" : "3x"
+    },
+    {
+      "idiom" : "iphone",
+      "size" : "29x29",
+      "scale" : "2x"
+    },
+    {
+      "idiom" : "iphone",
+      "size" : "29x29",
+      "scale" : "3x"
+    },
+    {
+      "idiom" : "iphone",
+      "size" : "40x40",
+      "scale" : "2x"
+    },
+    {
+      "idiom" : "iphone",
+      "size" : "40x40",
+      "scale" : "3x"
+    },
+    {
+      "idiom" : "iphone",
+      "size" : "60x60",
+      "scale" : "2x"
+    },
+    {
+      "idiom" : "iphone",
+      "size" : "60x60",
+      "scale" : "3x"
+    },
+    {
+      "idiom" : "ipad",
+      "size" : "20x20",
+      "scale" : "1x"
+    },
+    {
+      "idiom" : "ipad",
+      "size" : "20x20",
+      "scale" : "2x"
+    },
+    {
+      "idiom" : "ipad",
+      "size" : "29x29",
+      "scale" : "1x"
+    },
+    {
+      "idiom" : "ipad",
+      "size" : "29x29",
+      "scale" : "2x"
+    },
+    {
+      "idiom" : "ipad",
+      "size" : "40x40",
+      "scale" : "1x"
+    },
+    {
+      "idiom" : "ipad",
+      "size" : "40x40",
+      "scale" : "2x"
+    },
+    {
+      "idiom" : "ipad",
+      "size" : "76x76",
+      "scale" : "1x"
+    },
+    {
+      "idiom" : "ipad",
+      "size" : "76x76",
+      "scale" : "2x"
+    },
+    {
+      "idiom" : "ipad",
+      "size" : "83.5x83.5",
+      "scale" : "2x"
+    },
+    {
+      "idiom" : "ios-marketing",
+      "size" : "1024x1024",
+      "scale" : "1x"
+    }
+  ],
+  "info" : {
+    "version" : 1,
+    "author" : "xcode"
+  }
+}
@@ -0,0 +1,27 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<document type="com.apple.InterfaceBuilder3.CocoaTouch.Storyboard.XIB" version="3.0" toolsVersion="11134" systemVersion="15F34" targetRuntime="iOS.CocoaTouch" propertyAccessControl="none" useAutolayout="YES" launchScreen="YES" useTraitCollections="YES" colorMatched="YES" initialViewController="01J-lp-oVM">
+    <dependencies>
+        <plugIn identifier="com.apple.InterfaceBuilder.IBCocoaTouchPlugin" version="11106"/>
+        <capability name="documents saved in the Xcode 8 format" minToolsVersion="8.0"/>
+    </dependencies>
+    <scenes>
+        <!--View Controller-->
+        <scene sceneID="EHf-IW-A2E">
+            <objects>
+                <viewController id="01J-lp-oVM" sceneMemberID="viewController">
+                    <layoutGuides>
+                        <viewControllerLayoutGuide type="top" id="Llm-lL-Icb"/>
+                        <viewControllerLayoutGuide type="bottom" id="xb3-aO-Qok"/>
+                    </layoutGuides>
+                    <view key="view" contentMode="scaleToFill" id="Ze5-6b-2t3">
+                        <rect key="frame" x="0.0" y="0.0" width="375" height="667"/>
+                        <autoresizingMask key="autoresizingMask" widthSizable="YES" heightSizable="YES"/>
+                        <color key="backgroundColor" red="1" green="1" blue="1" alpha="1" colorSpace="custom" customColorSpace="sRGB"/>
+                    </view>
+                </viewController>
+                <placeholder placeholderIdentifier="IBFirstResponder" id="iYj-Kq-Ea1" userLabel="First Responder" sceneMemberID="firstResponder"/>
+            </objects>
+            <point key="canvasLocation" x="53" y="375"/>
+        </scene>
+    </scenes>
+</document>
@@ -0,0 +1,32 @@
+<?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">
+        <adaptation id="fullscreen"/>
+    </device>
+    <dependencies>
+        <deployment identifier="iOS"/>
+        <plugIn identifier="com.apple.InterfaceBuilder.IBCocoaTouchPlugin" version="13772"/>
+        <capability name="documents saved in the Xcode 8 format" minToolsVersion="8.0"/>
+    </dependencies>
+    <scenes>
+        <!--View Controller-->
+        <scene sceneID="tne-QT-ifu">
+            <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"/>
+                        <color key="backgroundColor" white="1" alpha="1" colorSpace="calibratedWhite"/>
+                        <fontDescription key="fontDescription" type="system" pointSize="14"/>
+                        <textInputTraits key="textInputTraits" autocapitalizationType="sentences"/>
+                    </textView>
+                    <connections>
+                        <outlet property="mainText" destination="CtX-mx-X98" id="oJy-5J-NKp"/>
+                    </connections>
+                </viewController>
+                <placeholder placeholderIdentifier="IBFirstResponder" id="dkx-z0-nzr" sceneMemberID="firstResponder"/>
+            </objects>
+            <point key="canvasLocation" x="32.799999999999997" y="32.833583208395808"/>
+        </scene>
+    </scenes>
+</document>
@@ -0,0 +1,45 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+	<key>CFBundleDevelopmentRegion</key>
+	<string>en</string>
+	<key>CFBundleExecutable</key>
+	<string>$(EXECUTABLE_NAME)</string>
+	<key>CFBundleIdentifier</key>
+	<string>$(PRODUCT_BUNDLE_IDENTIFIER)</string>
+	<key>CFBundleInfoDictionaryVersion</key>
+	<string>6.0</string>
+	<key>CFBundleName</key>
+	<string>$(PRODUCT_NAME)</string>
+	<key>CFBundlePackageType</key>
+	<string>APPL</string>
+	<key>CFBundleShortVersionString</key>
+	<string>1.0</string>
+	<key>CFBundleVersion</key>
+	<string>1</string>
+	<key>LSRequiresIPhoneOS</key>
+	<true/>
+	<key>UILaunchStoryboardName</key>
+	<string>LaunchScreen</string>
+	<key>UIMainStoryboardFile</key>
+	<string>Main</string>
+	<key>UIRequiredDeviceCapabilities</key>
+	<array>
+		<string>armv7</string>
+	</array>
+	<key>UISupportedInterfaceOrientations</key>
+	<array>
+		<string>UIInterfaceOrientationPortrait</string>
+		<string>UIInterfaceOrientationLandscapeLeft</string>
+		<string>UIInterfaceOrientationLandscapeRight</string>
+	</array>
+	<key>UISupportedInterfaceOrientations~ipad</key>
+	<array>
+		<string>UIInterfaceOrientationPortrait</string>
+		<string>UIInterfaceOrientationPortraitUpsideDown</string>
+		<string>UIInterfaceOrientationLandscapeLeft</string>
+		<string>UIInterfaceOrientationLandscapeRight</string>
+	</array>
+</dict>
+</plist>
@@ -0,0 +1,34 @@
+//
+//  ViewController.swift
+//  VoskApiTest
+//
+//  Created by Niсkolay Shmyrev on 01.03.20.
+//  Copyright © 2020 Alpha Cephei. All rights reserved.
+//
+
+import UIKit
+
+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..."
+            }
+            let vosk = Vosk()
+            let res = vosk.recognizeFile()
+            DispatchQueue.main.async {
+                self.mainText.text = res
+            }
+        }
+    }
+
+    override func didReceiveMemoryWarning() {
+        super.didReceiveMemoryWarning()
+    }
+}
+
@@ -0,0 +1,37 @@
+//
+//  Vosk.swift
+//  VoskApiTest
+//
+//  Created by Niсkolay Shmyrev on 01.03.20.
+//  Copyright © 2020 Alpha Cephei. All rights reserved.
+//
+
+import Foundation
+
+public final class Vosk {
+    
+    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);
+            }
+
+            vosk_recognizer_free(recognizer)
+            vosk_model_free(model)
+        }
+        return sres
+    }
+}
@@ -0,0 +1,48 @@
+// Copyright 2020 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.
+
+
+#ifndef _VOSK_API_H_
+#define _VOSK_API_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef struct VoskModel VoskModel;
+typedef struct VoskSpkModel VoskSpkModel;
+typedef struct VoskRecognizer VoskRecognizer;
+
+VoskModel *vosk_model_new(const char *model_path);
+void vosk_model_free(VoskModel *model);
+
+VoskSpkModel *vosk_spk_model_new(const char *model_path);
+void vosk_spk_model_free(VoskSpkModel *model);
+
+VoskRecognizer *vosk_recognizer_new(VoskModel *model, float sample_rate);
+VoskRecognizer *vosk_recognizer_new_spk(VoskModel *model, VoskSpkModel *spk_model, float sample_rate);
+VoskRecognizer *vosk_recognizer_new_grm(VoskModel *model, float sample_rate, const char *grammar);
+int vosk_recognizer_accept_waveform(VoskRecognizer *recognizer, const char *data, int length);
+int vosk_recognizer_accept_waveform_s(VoskRecognizer *recognizer, const short *data, int length);
+int vosk_recognizer_accept_waveform_f(VoskRecognizer *recognizer, const float *data, int length);
+const char *vosk_recognizer_result(VoskRecognizer *recognizer);
+const char *vosk_recognizer_partial_result(VoskRecognizer *recognizer);
+const char *vosk_recognizer_final_result(VoskRecognizer *recognizer);
+void vosk_recognizer_free(VoskRecognizer *recognizer);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif /* _VOSK_API_H_ */
@@ -0,0 +1 @@
+#import "Vosk/vosk_api.h"
@@ -11,6 +11,7 @@ KALDI_LIBS = \
             ${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 \
@@ -32,14 +33,16 @@ VOSK_SOURCES = \
 	../src/model.cc \
 	../src/model.h \
 	../src/spk_model.cc \
-	../src/spk_model.h
+	../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 -I../src -c++ \
+	swig -c++ -I../src \
 		-java -package org.kaldi \
 		-outdir org/kaldi -o $@ $<

@@ -47,11 +50,16 @@ clean:
 	$(RM) *.so *_wrap.cc *_wrap.o test/*.class
 	$(RM) -r org model-en

-model-en:
-	wget https://github.com/alphacep/kaldi-android-demo/releases/download/2020-01/alphacep-model-android-en-us-0.3.tar.gz
-	tar xf alphacep-model-android-en-us-0.3.tar.gz && rm alphacep-model-android-en-us-0.3.tar.gz
-	mv alphacep-model-android-en-us-0.3 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

-run: model-en
+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
@@ -0,0 +1,19 @@
+Java API sample
+
+Doesn't work on Windows or Mac yet, help to prepare the packaged jars is welcome.
+
+For now to try it:
+
+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
@@ -11,6 +11,7 @@ import java.nio.*;
 import org.kaldi.KaldiRecognizer;
 import org.kaldi.Model;
 import org.kaldi.SpkModel;
+import org.kaldi.Vosk;

 public class DecoderTest {
    static {
@@ -18,15 +19,17 @@ public class DecoderTest {
    }

    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-en");
+        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, nbytes)) {
+            if (rec.AcceptWaveform(b)) {
                System.out.println(rec.Result());
            } else {
                System.out.println(rec.PartialResult());
@@ -3,27 +3,27 @@ project(vosk)

 set(TOP_SRCDIR "${CMAKE_SOURCE_DIR}/..")
 if("x$ENV{WHEEL_FLAGS}" STREQUAL "x")
-    find_package (Python COMPONENTS Interpreter Development)
+    find_package (Python3 COMPONENTS Interpreter Development)
 else()
    # docker case
-    set(Python_INCLUDE_DIR "")
-    set(Python_LIBRARY "")
+    set(Python3_INCLUDE_DIRS "")
    set(TOP_SRCDIR "/io")
 endif()

 set(KALDI_ROOT "$ENV{KALDI_ROOT}")
-set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11 -O3 -DFST_NO_DYNAMIC_LINKING")
-include_directories("${TOP_SRCDIR}/src" "${KALDI_ROOT}/src" "${KALDI_ROOT}/tools/openfst/include" ${Python_INCLUDE_DIR})
-set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} $ENV{WHEEL_FLAGS}")
+set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DFST_NO_DYNAMIC_LINKING")
+set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} $ENV{WHEEL_FLAGS}")
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${CMAKE_C_FLAGS} -std=c++11")
+include_directories("${TOP_SRCDIR}/src" "${KALDI_ROOT}/src" "${KALDI_ROOT}/tools/openfst/include" ${Python3_INCLUDE_DIRS})

 find_package(SWIG REQUIRED)
 include(${SWIG_USE_FILE})

-set_property(SOURCE "${TOP_SRCDIR}/src/vosk.i" PROPERTY CPLUSPLUS ON)
 swig_add_library(vosk TYPE SHARED LANGUAGE Python OUTPUT_DIR "${CMAKE_LIBRARY_OUTPUT_DIRECTORY}" OUTFILE_DIR "."
-    SOURCES "${TOP_SRCDIR}/src/kaldi_recognizer.cc" 
+    SOURCES "${TOP_SRCDIR}/src/kaldi_recognizer.cc"
            "${TOP_SRCDIR}/src/spk_model.cc"
            "${TOP_SRCDIR}/src/model.cc"
+            "${TOP_SRCDIR}/src/vosk_api.cc"
            "${TOP_SRCDIR}/src/vosk.i")

 swig_link_libraries(vosk
@@ -35,6 +35,7 @@ swig_link_libraries(vosk
             ${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
@@ -45,6 +46,6 @@ swig_link_libraries(vosk
             ${KALDI_ROOT}/tools/openfst/lib/libfst.a
             ${KALDI_ROOT}/tools/openfst/lib/libfstngram.a
             ${KALDI_ROOT}/tools/OpenBLAS/libopenblas.a
-             -lgfortran)
+             -lgfortran -lstdc++)

 set_target_properties(_vosk PROPERTIES LINK_FLAGS_RELEASE -s)
@@ -4,7 +4,7 @@ from vosk import Model, KaldiRecognizer
 import sys
 import json

-model = Model("model-en")
+model = Model("model")
 rec = KaldiRecognizer(model, 8000)

 res = json.loads(rec.FinalResult())
@@ -3,8 +3,8 @@
 from vosk import Model, KaldiRecognizer
 import os

-if not os.path.exists("model-en"):
-    print ("Please download the model from https://github.com/alphacep/kaldi-android-demo/releases and unpack as 'model' in the current folder.")
+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)

 import pyaudio
@@ -13,11 +13,11 @@ p = pyaudio.PyAudio()
 stream = p.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True, frames_per_buffer=8000)
 stream.start_stream()

-model = Model("model-en")
+model = Model("model")
 rec = KaldiRecognizer(model, 16000)

 while True:
-    data = stream.read(2000)
+    data = stream.read(4000)
    if len(data) == 0:
        break
    if rec.AcceptWaveform(data):
@@ -1,12 +1,14 @@
 #!/usr/bin/python3

-from vosk import Model, KaldiRecognizer
+from vosk import Model, KaldiRecognizer, SetLogLevel
 import sys
 import os
 import wave

-if not os.path.exists("model-en"):
-    print ("Please download the model from https://github.com/alphacep/kaldi-android-demo/releases and unpack as 'model-en' in the current folder.")
+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.")
    exit (1)

 wf = wave.open(sys.argv[1], "rb")
@@ -14,11 +16,11 @@ 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-en")
+model = Model("model")
 rec = KaldiRecognizer(model, wf.getframerate())

 while True:
-    data = wf.readframes(1000)
+    data = wf.readframes(4000)
    if len(data) == 0:
        break
    if rec.AcceptWaveform(data):
@@ -7,15 +7,15 @@ import json
 import os
 import numpy as np

-model_path = "model-en"
+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/kaldi-android-demo/releases and unpack as {} in the current folder.".format(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))
    exit (1)

 if not os.path.exists(spk_model_path):
-    print ("Please download the speaker model from https://github.com/alphacep/kaldi-android-demo/releases and unpack as {} in the current folder.".format(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))
    exit (1)

 wf = wave.open(sys.argv[1], "rb")
@@ -38,7 +38,7 @@ def cosine_dist(x, y):
    return 1 - np.dot(nx, ny) / np.linalg.norm(nx) / np.linalg.norm(ny)

 while True:
-    data = wf.readframes(1000)
+    data = wf.readframes(4000)
    if len(data) == 0:
        break
    if rec.AcceptWaveform(data):
@@ -5,12 +5,12 @@ import sys
 import json
 import os

-if not os.path.exists("model-en"):
-    print ("Please download the model from https://github.com/alphacep/kaldi-android-demo/releases and unpack as 'model-en' in the current folder.")
+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)


-model = Model("model-en")
+model = Model("model")

 # Large vocabulary free form recognition
 rec = KaldiRecognizer(model, 16000)
@@ -22,7 +22,7 @@ wf = open(sys.argv[1], "rb")
 wf.read(44) # skip header

 while True:
-    data = wf.read(2000)
+    data = wf.read(4000)
    if len(data) == 0:
        break
    if rec.AcceptWaveform(data):
@@ -5,8 +5,8 @@ import sys
 import os
 import wave

-if not os.path.exists("model-en"):
-    print ("Please download the model from https://github.com/alphacep/kaldi-android-demo/releases and unpack as 'model-en' in the current folder.")
+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)

 wf = wave.open(sys.argv[1], "rb")
@@ -14,12 +14,12 @@ 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-en")
+model = Model("model")
 # You can also specify the possible word list
 rec = KaldiRecognizer(model, wf.getframerate(), "zero oh one two three four five six seven eight nine")

 while True:
-    data = wf.readframes(1000)
+    data = wf.readframes(4000)
    if len(data) == 0:
        break
    if rec.AcceptWaveform(data):
@@ -7,7 +7,7 @@ with open("README.md", "r") as fh:

 setuptools.setup(
    name="vosk", # Replace with your own username
-    version="0.3.3",
+    version="0.3.7",
    author="Alpha Cephei Inc",
    author_email="contact@alphacephei.com",
    description="API for Kaldi and Vosk",
@@ -0,0 +1,27 @@
+#!/usr/bin/python3
+
+from vosk import Model, KaldiRecognizer
+import sys
+import os
+import wave
+import json
+
+model = Model("model")
+
+for line in open(sys.argv[1]):
+    uid, fn = line.split()
+    wf = wave.open(fn, "rb")
+    rec = KaldiRecognizer(model, wf.getframerate())
+
+    text = ""
+    while True:
+        data = wf.readframes(1000)
+        if len(data) == 0:
+            break
+        if rec.AcceptWaveform(data):
+            jres = json.loads(rec.Result())
+            text = text + " " + jres['text']
+    jres = json.loads(rec.FinalResult())
+    text = text + " " + jres['text']
+    print (uid + text)
+
@@ -1 +1,2 @@
-from .vosk import KaldiRecognizer, Model, SpkModel
+from .vosk import KaldiRecognizer, Model, SpkModel, SetLogLevel
+
@@ -20,110 +20,144 @@
 using namespace fst;
 using namespace kaldi::nnet3;

-KaldiRecognizer::KaldiRecognizer(Model &model, float sample_frequency) : model_(model), spk_model_(0), sample_frequency_(sample_frequency) {
+KaldiRecognizer::KaldiRecognizer(Model *model, float sample_frequency) : model_(model), spk_model_(0), sample_frequency_(sample_frequency) {

-    feature_pipeline_ = new kaldi::OnlineNnet2FeaturePipeline (model_.feature_info_);
-    silence_weighting_ = new kaldi::OnlineSilenceWeighting(*model_.trans_model_, model_.feature_info_.silence_weighting_config, 3);
+    model_->Ref();

+    feature_pipeline_ = new kaldi::OnlineNnet2FeaturePipeline (model_->feature_info_);
+    silence_weighting_ = new kaldi::OnlineSilenceWeighting(*model_->trans_model_, model_->feature_info_.silence_weighting_config, 3);
+
+    g_fst_ = NULL;
    decode_fst_ = NULL;

-    if (!model_.hclg_fst_) {
-        if (model_.hcl_fst_ && model_.g_fst_) {
-            decode_fst_ = LookaheadComposeFst(*model_.hcl_fst_, *model_.g_fst_, model_.disambig_);
+    if (!model_->hclg_fst_) {
+        if (model_->hcl_fst_ && model_->g_fst_) {
+            decode_fst_ = LookaheadComposeFst(*model_->hcl_fst_, *model_->g_fst_, model_->disambig_);
        } else {
            KALDI_ERR << "Can't create decoding graph";
        }
    }

-    decoder_ = new kaldi::SingleUtteranceNnet3Decoder(model_.nnet3_decoding_config_,
-            *model_.trans_model_,
-            *model_.decodable_info_,
-            model_.hclg_fst_ ? *model.hclg_fst_ : *decode_fst_,
+    decoder_ = new kaldi::SingleUtteranceNnet3Decoder(model_->nnet3_decoding_config_,
+            *model_->trans_model_,
+            *model_->decodable_info_,
+            model_->hclg_fst_ ? *model_->hclg_fst_ : *decode_fst_,
            feature_pipeline_);

    frame_offset_ = 0;
    input_finalized_ = false;
    spk_feature_ = NULL;
+
+    InitRescoring();
 }

-KaldiRecognizer::KaldiRecognizer(Model &model, float sample_frequency, char const *grammar) : model_(model), spk_model_(0), sample_frequency_(sample_frequency)
+KaldiRecognizer::KaldiRecognizer(Model *model, float sample_frequency, char const *grammar) : model_(model), spk_model_(0), sample_frequency_(sample_frequency)
 {
-    feature_pipeline_ = new kaldi::OnlineNnet2FeaturePipeline (model_.feature_info_);
-    silence_weighting_ = new kaldi::OnlineSilenceWeighting(*model_.trans_model_, model_.feature_info_.silence_weighting_config, 3);
+    model_->Ref();

-    if (model_.hcl_fst_) {
-        g_fst_.AddState();
-        g_fst_.SetStart(0);
-        g_fst_.AddState();
-        g_fst_.SetFinal(1, fst::TropicalWeight::One());
-        g_fst_.AddArc(1, StdArc(0, 0, fst::TropicalWeight::One(), 0));
+    feature_pipeline_ = new kaldi::OnlineNnet2FeaturePipeline (model_->feature_info_);
+    silence_weighting_ = new kaldi::OnlineSilenceWeighting(*model_->trans_model_, model_->feature_info_.silence_weighting_config, 3);
+
+    g_fst_ = new StdVectorFst();
+    if (model_->hcl_fst_) {
+        g_fst_->AddState();
+        g_fst_->SetStart(0);
+        g_fst_->AddState();
+        g_fst_->SetFinal(1, fst::TropicalWeight::One());
+        g_fst_->AddArc(1, StdArc(0, 0, fst::TropicalWeight::One(), 0));

        // Create simple word loop FST
-        std::stringstream ss(grammar);
-        std::string token;
+        stringstream ss(grammar);
+        string token;

-        while (std::getline(ss, token, ' ')) {
-            int32 id = model_.word_syms_->Find(token);
-            g_fst_.AddArc(0, StdArc(id, id, fst::TropicalWeight::One(), 1));
+        while (getline(ss, token, ' ')) {
+            int32 id = model_->word_syms_->Find(token);
+            g_fst_->AddArc(0, StdArc(id, id, fst::TropicalWeight::One(), 1));
        }
-        ArcSort(&g_fst_, ILabelCompare<StdArc>());
+        ArcSort(g_fst_, ILabelCompare<StdArc>());

-        decode_fst_ = LookaheadComposeFst(*model_.hcl_fst_, g_fst_, model_.disambig_);
+        decode_fst_ = LookaheadComposeFst(*model_->hcl_fst_, *g_fst_, model_->disambig_);
    } else {
        decode_fst_ = NULL;
        KALDI_ERR << "Can't create decoding graph";
    }

-    decoder_ = new kaldi::SingleUtteranceNnet3Decoder(model_.nnet3_decoding_config_,
-            *model_.trans_model_,
-            *model_.decodable_info_,
-            model_.hclg_fst_ ? *model.hclg_fst_ : *decode_fst_,
+    decoder_ = new kaldi::SingleUtteranceNnet3Decoder(model_->nnet3_decoding_config_,
+            *model_->trans_model_,
+            *model_->decodable_info_,
+            model_->hclg_fst_ ? *model_->hclg_fst_ : *decode_fst_,
            feature_pipeline_);

    frame_offset_ = 0;
    input_finalized_ = false;
    spk_feature_ = NULL;
+
+    InitRescoring();
 }

+KaldiRecognizer::KaldiRecognizer(Model *model, SpkModel *spk_model, float sample_frequency) : model_(model), spk_model_(spk_model), sample_frequency_(sample_frequency) {

-KaldiRecognizer::KaldiRecognizer(Model &model, SpkModel *spk_model, float sample_frequency) : model_(model), spk_model_(spk_model), sample_frequency_(sample_frequency) {
-    feature_pipeline_ = new kaldi::OnlineNnet2FeaturePipeline (model_.feature_info_);
-    silence_weighting_ = new kaldi::OnlineSilenceWeighting(*model_.trans_model_, model_.feature_info_.silence_weighting_config, 3);
+    model_->Ref();
+    spk_model->Ref();
+
+    feature_pipeline_ = new kaldi::OnlineNnet2FeaturePipeline (model_->feature_info_);
+    silence_weighting_ = new kaldi::OnlineSilenceWeighting(*model_->trans_model_, model_->feature_info_.silence_weighting_config, 3);

    decode_fst_ = NULL;
+    g_fst_ = NULL;

-    if (!model_.hclg_fst_) {
-        if (model_.hcl_fst_ && model_.g_fst_) {
-            decode_fst_ = LookaheadComposeFst(*model_.hcl_fst_, *model_.g_fst_, model_.disambig_);
+    if (!model_->hclg_fst_) {
+        if (model_->hcl_fst_ && model_->g_fst_) {
+            decode_fst_ = LookaheadComposeFst(*model_->hcl_fst_, *model_->g_fst_, model_->disambig_);
        } else {
            KALDI_ERR << "Can't create decoding graph";
        }
    }

-    decoder_ = new kaldi::SingleUtteranceNnet3Decoder(model_.nnet3_decoding_config_,
-            *model_.trans_model_,
-            *model_.decodable_info_,
-            model_.hclg_fst_ ? *model.hclg_fst_ : *decode_fst_,
+    decoder_ = new kaldi::SingleUtteranceNnet3Decoder(model_->nnet3_decoding_config_,
+            *model_->trans_model_,
+            *model_->decodable_info_,
+            model_->hclg_fst_ ? *model_->hclg_fst_ : *decode_fst_,
            feature_pipeline_);

    frame_offset_ = 0;
    input_finalized_ = false;

    spk_feature_ = new OnlineMfcc(spk_model_->spkvector_mfcc_opts);
+
+    InitRescoring();
 }

 KaldiRecognizer::~KaldiRecognizer() {
    delete feature_pipeline_;
    delete silence_weighting_;
    delete decoder_;
+    delete g_fst_;
    delete decode_fst_;
    delete spk_feature_;
+    delete lm_fst_;
+
+    model_->Unref();
+    if (spk_model_)
+         spk_model_->Unref();
+}
+
+void KaldiRecognizer::InitRescoring()
+{
+    if (model_->std_lm_fst_) {
+        fst::CacheOptions cache_opts(true, 50000);
+        fst::MapFstOptions mapfst_opts(cache_opts);
+        fst::StdToLatticeMapper<kaldi::BaseFloat> mapper;
+        lm_fst_ = new fst::MapFst<fst::StdArc, kaldi::LatticeArc, fst::StdToLatticeMapper<kaldi::BaseFloat> >(*model_->std_lm_fst_, mapper, mapfst_opts);
+    } else {
+        lm_fst_ = NULL;
+    }
 }

 void KaldiRecognizer::CleanUp()
 {
    delete silence_weighting_;
-    silence_weighting_ = new kaldi::OnlineSilenceWeighting(*model_.trans_model_, model_.feature_info_.silence_weighting_config, 3);
+    silence_weighting_ = new kaldi::OnlineSilenceWeighting(*model_->trans_model_, model_->feature_info_.silence_weighting_config, 3);

    frame_offset_ += decoder_->NumFramesDecoded();
    decoder_->InitDecoding(frame_offset_);
@@ -133,7 +167,7 @@ void KaldiRecognizer::UpdateSilenceWeights()
 {
    if (silence_weighting_->Active() && feature_pipeline_->NumFramesReady() > 0 &&
        feature_pipeline_->IvectorFeature() != NULL) {
-        std::vector<std::pair<int32, BaseFloat> > delta_weights;
+        vector<pair<int32, BaseFloat> > delta_weights;
        silence_weighting_->ComputeCurrentTraceback(decoder_->Decoder());
        silence_weighting_->GetDeltaWeights(feature_pipeline_->NumFramesReady(),
                                          frame_offset_ * 3,
@@ -184,14 +218,13 @@ bool KaldiRecognizer::AcceptWaveform(Vector<BaseFloat> &wdata)
        spk_feature_->AcceptWaveform(sample_frequency_, wdata);
    }

-    if (decoder_->EndpointDetected(model_.endpoint_config_)) {
+    if (decoder_->EndpointDetected(model_->endpoint_config_)) {
        return true;
    }

    return false;
 }

-
 // Computes an xvector from a chunk of speech features.
 static void RunNnetComputation(const MatrixBase<BaseFloat> &features,
    const nnet3::Nnet &nnet, nnet3::CachingOptimizingCompiler *compiler,
@@ -208,7 +241,7 @@ static void RunNnetComputation(const MatrixBase<BaseFloat> &features,
    output_spec.indexes.resize(1);
    request.outputs.resize(1);
    request.outputs[0].Swap(&output_spec);
-    std::shared_ptr<const nnet3::NnetComputation> computation = compiler->Compile(request);
+    shared_ptr<const nnet3::NnetComputation> computation = compiler->Compile(request);
    nnet3::Nnet *nnet_to_update = NULL;  // we're not doing any update.
    nnet3::NnetComputer computer(nnet3::NnetComputeOptions(), *computation,
                    nnet, nnet_to_update);
@@ -221,7 +254,6 @@ static void RunNnetComputation(const MatrixBase<BaseFloat> &features,
    xvector->CopyFromVec(cu_output.Row(0));
 }

-
 void KaldiRecognizer::GetSpkVector(Vector<BaseFloat> &xvector)
 {
    int num_frames = spk_feature_->NumFramesReady() - frame_offset_ * 3;
@@ -242,8 +274,7 @@ void KaldiRecognizer::GetSpkVector(Vector<BaseFloat> &xvector)
    RunNnetComputation(features, spk_model_->speaker_nnet, &compiler, &xvector);
 }

-
-std::string KaldiRecognizer::Result()
+const char* KaldiRecognizer::Result()
 {

    if (!input_finalized_) {
@@ -252,35 +283,59 @@ std::string KaldiRecognizer::Result()
    }

    if (decoder_->NumFramesDecoded() == 0) {
-        return "{\"text\": \"\"}";
+        last_result_ = "{\"text\": \"\"}";
+        return last_result_.c_str();
    }

    kaldi::CompactLattice clat;
    decoder_->GetLattice(true, &clat);
-    fst::ScaleLattice(fst::LatticeScale(8.0, 10.0), &clat);

+    if (model_->std_lm_fst_) {
+        Lattice lat1;
+
+        ConvertLattice(clat, &lat1);
+        fst::ScaleLattice(fst::GraphLatticeScale(-1.0), &lat1);
+        fst::ArcSort(&lat1, fst::OLabelCompare<kaldi::LatticeArc>());
+        kaldi::Lattice composed_lat;
+        fst::Compose(lat1, *lm_fst_, &composed_lat);
+        fst::Invert(&composed_lat);
+        kaldi::CompactLattice determinized_lat;
+        DeterminizeLattice(composed_lat, &determinized_lat);
+        fst::ScaleLattice(fst::GraphLatticeScale(-1), &determinized_lat);
+        fst::ArcSort(&determinized_lat, fst::OLabelCompare<kaldi::CompactLatticeArc>());
+
+        kaldi::ConstArpaLmDeterministicFst const_arpa_fst(model_->const_arpa_);
+        kaldi::CompactLattice composed_clat;
+        kaldi::ComposeCompactLatticeDeterministic(determinized_lat, &const_arpa_fst, &composed_clat);
+        kaldi::Lattice composed_lat1;
+        ConvertLattice(composed_clat, &composed_lat1);
+        fst::Invert(&composed_lat1);
+        DeterminizeLattice(composed_lat1, &clat);
+    }
+
+    fst::ScaleLattice(fst::LatticeScale(9.0, 10.0), &clat);
    CompactLattice aligned_lat;
-    if (model_.winfo_) {
-        WordAlignLattice(clat, *model_.trans_model_, *model_.winfo_, 0, &aligned_lat);
+    if (model_->winfo_) {
+        WordAlignLattice(clat, *model_->trans_model_, *model_->winfo_, 0, &aligned_lat);
    } else {
        aligned_lat = clat;
    }

    MinimumBayesRisk mbr(aligned_lat);
-    const std::vector<BaseFloat> &conf = mbr.GetOneBestConfidences();
-    const std::vector<int32> &words = mbr.GetOneBest();
-    const std::vector<std::pair<BaseFloat, BaseFloat> > &times =
+    const vector<BaseFloat> &conf = mbr.GetOneBestConfidences();
+    const vector<int32> &words = mbr.GetOneBest();
+    const vector<pair<BaseFloat, BaseFloat> > &times =
          mbr.GetOneBestTimes();

    int size = words.size();

    json::JSON obj;
-    std::stringstream text;
+    stringstream text;

    // Create JSON object
    for (int i = 0; i < size; i++) {
        json::JSON word;
-        word["word"] = model_.word_syms_->Find(words[i]);
+        word["word"] = model_->word_syms_->Find(words[i]);
        word["start"] = (frame_offset_ + times[i].first) * 0.03;
        word["end"] = (frame_offset_ + times[i].second) * 0.03;
        word["conf"] = conf[i];
@@ -289,7 +344,7 @@ std::string KaldiRecognizer::Result()
        if (i) {
            text << " ";
        }
-        text << model_.word_syms_->Find(words[i]);
+        text << model_->word_syms_->Find(words[i]);
    }
    obj["text"] = text.str();

@@ -301,36 +356,39 @@ std::string KaldiRecognizer::Result()
        }
    }

-    return obj.dump();
+    last_result_ = obj.dump();
+    return last_result_.c_str();
 }

-std::string KaldiRecognizer::PartialResult()
+const char* KaldiRecognizer::PartialResult()
 {
    json::JSON res;
    if (decoder_->NumFramesDecoded() == 0) {
        res["partial"] = "";
-        return res.dump();
+        last_result_ = res.dump();
+        return last_result_.c_str();
    }

    kaldi::Lattice lat;
    decoder_->GetBestPath(false, &lat);
-    std::vector<kaldi::int32> alignment, words;
+    vector<kaldi::int32> alignment, words;
    LatticeWeight weight;
    GetLinearSymbolSequence(lat, &alignment, &words, &weight);

-    std::ostringstream text;
+    ostringstream text;
    for (size_t i = 0; i < words.size(); i++) {
        if (i) {
            text << " ";
        }
-        text << model_.word_syms_->Find(words[i]);
+        text << model_->word_syms_->Find(words[i]);
    }
    res["partial"] = text.str();

-    return res.dump();
+    last_result_ = res.dump();
+    return last_result_.c_str();
 }

-std::string KaldiRecognizer::FinalResult()
+const char* KaldiRecognizer::FinalResult()
 {
    if (!input_finalized_) {
        feature_pipeline_->InputFinished();
@@ -338,6 +396,9 @@ std::string KaldiRecognizer::FinalResult()
        decoder_->AdvanceDecoding();
        decoder_->FinalizeDecoding();
        input_finalized_ = true;
+        return Result();
+    } else {
+        last_result_ = "{\"text\": \"\"}";
+        return last_result_.c_str();
    }
-    return Result();
 }
@@ -31,34 +31,38 @@ using namespace kaldi;

 class KaldiRecognizer {
    public:
-        KaldiRecognizer(Model &model, float sample_frequency);
-        KaldiRecognizer(Model &model, SpkModel *spk_model, float sample_frequency);
-        KaldiRecognizer(Model &model, float sample_frequency, char const *grammar);
+        KaldiRecognizer(Model *model, float sample_frequency);
+        KaldiRecognizer(Model *model, SpkModel *spk_model, float sample_frequency);
+        KaldiRecognizer(Model *model, float sample_frequency, char const *grammar);
        ~KaldiRecognizer();
        bool AcceptWaveform(const char *data, int len);
        bool AcceptWaveform(const short *sdata, int len);
        bool AcceptWaveform(const float *fdata, int len);
-        std::string Result();
-        std::string FinalResult();
-        std::string PartialResult();
+        const char* Result();
+        const char* FinalResult();
+        const char* PartialResult();

    private:
+        void InitRescoring();
        void CleanUp();
        void UpdateSilenceWeights();
        bool AcceptWaveform(Vector<BaseFloat> &wdata);
        void GetSpkVector(Vector<BaseFloat> &xvector);

-        Model &model_;
+        Model *model_;
        SingleUtteranceNnet3Decoder *decoder_;
        fst::LookaheadFst<fst::StdArc, int32> *decode_fst_;
-        fst::StdVectorFst g_fst_; // dynamically constructed grammar
+        fst::StdVectorFst *g_fst_; // dynamically constructed grammar
        OnlineNnet2FeaturePipeline *feature_pipeline_;
        OnlineSilenceWeighting *silence_weighting_;

        SpkModel *spk_model_;
        OnlineBaseFeature *spk_feature_;

+        fst::MapFst<fst::StdArc, kaldi::LatticeArc, fst::StdToLatticeMapper<kaldi::BaseFloat> > *lm_fst_;
+
        float sample_frequency_;
        int32 frame_offset_;
        bool input_finalized_;
+        string last_result_;
 };
@@ -45,19 +45,97 @@ static FstRegisterer<NGramFst<StdArc>> NGramFst_StdArc_registerer;

 #ifdef __ANDROID__
 #include <android/log.h>
-static void AndroidLogHandler(const LogMessageEnvelope &env, const char *message)
+static void KaldiLogHandler(const LogMessageEnvelope &env, const char *message)
 {
-    __android_log_print(ANDROID_LOG_VERBOSE, "KaldiDemo", message, 1);
+  int priority;
+  if (env.severity > GetVerboseLevel())
+      return;
+
+  if (env.severity > LogMessageEnvelope::kInfo) {
+    priority = ANDROID_LOG_VERBOSE;
+  } else {
+    switch (env.severity) {
+    case LogMessageEnvelope::kInfo:
+      priority = ANDROID_LOG_INFO;
+      break;
+    case LogMessageEnvelope::kWarning:
+      priority = ANDROID_LOG_WARN;
+      break;
+    case LogMessageEnvelope::kAssertFailed:
+      priority = ANDROID_LOG_FATAL;
+      break;
+    case LogMessageEnvelope::kError:
+    default: // If not the ERROR, it still an error!
+      priority = ANDROID_LOG_ERROR;
+      break;
+    }
+  }
+
+  std::stringstream full_message;
+  full_message << env.func << "():" << env.file << ':'
+               << env.line << ") " << message;
+
+  __android_log_print(priority, "VoskAPI", "%s", full_message.str().c_str());
+}
+#else
+static void KaldiLogHandler(const LogMessageEnvelope &env, const char *message)
+{
+  if (env.severity > GetVerboseLevel())
+      return;
+
+  // Modified default Kaldi logging so we can disable LOG messages.
+  std::stringstream full_message;
+  if (env.severity > LogMessageEnvelope::kInfo) {
+    full_message << "VLOG[" << env.severity << "] (";
+  } else {
+    switch (env.severity) {
+    case LogMessageEnvelope::kInfo:
+      full_message << "LOG (";
+      break;
+    case LogMessageEnvelope::kWarning:
+      full_message << "WARNING (";
+      break;
+    case LogMessageEnvelope::kAssertFailed:
+      full_message << "ASSERTION_FAILED (";
+      break;
+    case LogMessageEnvelope::kError:
+    default: // If not the ERROR, it still an error!
+      full_message << "ERROR (";
+      break;
+    }
+  }
+  // Add other info from the envelope and the message text.
+  full_message << "VoskAPI" << ':'
+               << env.func << "():" << env.file << ':'
+               << env.line << ") " << message;
+
+  // Print the complete message to stderr.
+  full_message << "\n";
+  std::cerr << full_message.str();
 }
 #endif

-Model::Model(const char *model_path) {
+Model::Model(const char *model_path) : model_path_str_(model_path) {

-#ifdef __ANDROID__
-    SetLogHandler(AndroidLogHandler);
-#endif
+    SetLogHandler(KaldiLogHandler);

-    const char *usage = "Read the docs";
+    struct stat buffer;
+    string am_path = model_path_str_ + "/am/final.mdl";
+    if (stat(am_path.c_str(), &buffer) == 0) {
+         ConfigureV2();
+    } else {
+         ConfigureV1();
+    }
+
+    ReadDataFiles();
+
+    ref_cnt_ = 1;
+}
+
+// Old model layout without model configuration file
+
+void Model::ConfigureV1()
+{
    const char *extra_args[] = {
        "--min-active=200",
        "--max-active=3000",
@@ -71,49 +149,92 @@ Model::Model(const char *model_path) {
        "--endpoint.rule2.min-trailing-silence=0.5",
        "--endpoint.rule3.min-trailing-silence=1.0",
        "--endpoint.rule4.min-trailing-silence=2.0",
-    };
-    std::string model_path_str(model_path);

-    kaldi::ParseOptions po(usage);
+        "--print-args=false",
+    };
+
+    kaldi::ParseOptions po("");
    nnet3_decoding_config_.Register(&po);
    endpoint_config_.Register(&po);
    decodable_opts_.Register(&po);

-    std::vector<const char*> args;
+    vector<const char*> args;
    args.push_back("vosk");
    args.insert(args.end(), extra_args, extra_args + sizeof(extra_args) / sizeof(extra_args[0]));
    po.Read(args.size(), args.data());

    feature_info_.feature_type = "mfcc";
-    ReadConfigFromFile(model_path_str + "/mfcc.conf", &feature_info_.mfcc_opts);
+    ReadConfigFromFile(model_path_str_ + "/mfcc.conf", &feature_info_.mfcc_opts);
    feature_info_.mfcc_opts.frame_opts.allow_downsample = true; // It is safe to downsample

    feature_info_.silence_weighting_config.silence_weight = 1e-3;
-    feature_info_.silence_weighting_config.silence_phones_str = "1:2:3:4:5:6:7:8:9:10";
+    feature_info_.silence_weighting_config.silence_phones_str = endpoint_config_.silence_phones;

    OnlineIvectorExtractionConfig ivector_extraction_opts;
-    ivector_extraction_opts.splice_config_rxfilename = model_path_str + "/ivector/splice.conf";
-    ivector_extraction_opts.cmvn_config_rxfilename = model_path_str + "/ivector/online_cmvn.conf";
-    ivector_extraction_opts.lda_mat_rxfilename = model_path_str + "/ivector/final.mat";
-    ivector_extraction_opts.global_cmvn_stats_rxfilename = model_path_str + "/ivector/global_cmvn.stats";
-    ivector_extraction_opts.diag_ubm_rxfilename = model_path_str + "/ivector/final.dubm";
-    ivector_extraction_opts.ivector_extractor_rxfilename = model_path_str + "/ivector/final.ie";
-    ivector_extraction_opts.num_gselect = 5;
-    ivector_extraction_opts.min_post = 0.025;
-    ivector_extraction_opts.posterior_scale = 0.1;
-    ivector_extraction_opts.max_remembered_frames = 1000;
-    ivector_extraction_opts.max_count = 100;
-    ivector_extraction_opts.ivector_period = 200;
+    ivector_extraction_opts.splice_config_rxfilename = model_path_str_ + "/ivector/splice.conf";
+    ivector_extraction_opts.cmvn_config_rxfilename = model_path_str_ + "/ivector/online_cmvn.conf";
+    ivector_extraction_opts.lda_mat_rxfilename = model_path_str_ + "/ivector/final.mat";
+    ivector_extraction_opts.global_cmvn_stats_rxfilename = model_path_str_ + "/ivector/global_cmvn.stats";
+    ivector_extraction_opts.diag_ubm_rxfilename = model_path_str_ + "/ivector/final.dubm";
+    ivector_extraction_opts.ivector_extractor_rxfilename = model_path_str_ + "/ivector/final.ie";
    feature_info_.use_ivectors = true;
    feature_info_.ivector_extractor_info.Init(ivector_extraction_opts);

-    nnet3_rxfilename_ = model_path_str + "/final.mdl";
-    hclg_fst_rxfilename_ = model_path_str + "/HCLG.fst";
-    hcl_fst_rxfilename_ = model_path_str + "/HCLr.fst";
-    g_fst_rxfilename_ = model_path_str + "/Gr.fst";
-    disambig_rxfilename_ = model_path_str + "/disambig_tid.int";
-    word_syms_rxfilename_ = model_path_str + "/words.txt";
-    winfo_rxfilename_ = model_path_str + "/word_boundary.int";
+    nnet3_rxfilename_ = model_path_str_ + "/final.mdl";
+    hclg_fst_rxfilename_ = model_path_str_ + "/HCLG.fst";
+    hcl_fst_rxfilename_ = model_path_str_ + "/HCLr.fst";
+    g_fst_rxfilename_ = model_path_str_ + "/Gr.fst";
+    disambig_rxfilename_ = model_path_str_ + "/disambig_tid.int";
+    word_syms_rxfilename_ = model_path_str_ + "/words.txt";
+    winfo_rxfilename_ = model_path_str_ + "/word_boundary.int";
+    carpa_rxfilename_ = model_path_str_ + "/rescore/G.carpa";
+    std_fst_rxfilename_ = model_path_str_ + "/rescore/G.fst";
+}
+
+void Model::ConfigureV2()
+{
+    kaldi::ParseOptions po("something");
+    nnet3_decoding_config_.Register(&po);
+    endpoint_config_.Register(&po);
+    decodable_opts_.Register(&po);
+    po.ReadConfigFile(model_path_str_ + "/conf/model.conf");
+
+    KALDI_LOG << "Decoding params beam=" << nnet3_decoding_config_.beam <<
+         " max-active=" << nnet3_decoding_config_.max_active <<
+         " lattice-beam=" << nnet3_decoding_config_.lattice_beam;
+    KALDI_LOG << "Silence phones " << endpoint_config_.silence_phones;
+
+    feature_info_.feature_type = "mfcc";
+    ReadConfigFromFile(model_path_str_ + "/conf/mfcc.conf", &feature_info_.mfcc_opts);
+    feature_info_.mfcc_opts.frame_opts.allow_downsample = true; // It is safe to downsample
+
+    feature_info_.silence_weighting_config.silence_weight = 1e-3;
+    feature_info_.silence_weighting_config.silence_phones_str = endpoint_config_.silence_phones;
+
+    OnlineIvectorExtractionConfig ivector_extraction_opts;
+    ivector_extraction_opts.splice_config_rxfilename = model_path_str_ + "/ivector/splice.conf";
+    ivector_extraction_opts.cmvn_config_rxfilename = model_path_str_ + "/ivector/online_cmvn.conf";
+    ivector_extraction_opts.lda_mat_rxfilename = model_path_str_ + "/ivector/final.mat";
+    ivector_extraction_opts.global_cmvn_stats_rxfilename = model_path_str_ + "/ivector/global_cmvn.stats";
+    ivector_extraction_opts.diag_ubm_rxfilename = model_path_str_ + "/ivector/final.dubm";
+    ivector_extraction_opts.ivector_extractor_rxfilename = model_path_str_ + "/ivector/final.ie";
+    feature_info_.use_ivectors = true;
+    feature_info_.ivector_extractor_info.Init(ivector_extraction_opts);
+
+    nnet3_rxfilename_ = model_path_str_ + "/am/final.mdl";
+    hclg_fst_rxfilename_ = model_path_str_ + "/graph/HCLG.fst";
+    hcl_fst_rxfilename_ = model_path_str_ + "/graph/HCLr.fst";
+    g_fst_rxfilename_ = model_path_str_ + "/graph/Gr.fst";
+    disambig_rxfilename_ = model_path_str_ + "/graph/disambig_tid.int";
+    word_syms_rxfilename_ = model_path_str_ + "/graph/words.txt";
+    winfo_rxfilename_ = model_path_str_ + "/graph/phones/word_boundary.int";
+    carpa_rxfilename_ = model_path_str_ + "/rescore/G.carpa";
+    std_fst_rxfilename_ = model_path_str_ + "/rescore/G.fst";
+}
+
+void Model::ReadDataFiles()
+{
+    struct stat buffer;

    trans_model_ = new kaldi::TransitionModel();
    nnet_ = new kaldi::nnet3::AmNnetSimple();
@@ -126,19 +247,19 @@ Model::Model(const char *model_path) {
        SetDropoutTestMode(true, &(nnet_->GetNnet()));
        nnet3::CollapseModel(nnet3::CollapseModelConfig(), &(nnet_->GetNnet()));
    }
-
    decodable_info_ = new nnet3::DecodableNnetSimpleLoopedInfo(decodable_opts_,
                                                               nnet_);
-    struct stat buffer;
+
    if (stat(hclg_fst_rxfilename_.c_str(), &buffer) == 0) {
+        KALDI_LOG << "Loading HCLG from " << hclg_fst_rxfilename_;
        hclg_fst_ = fst::ReadFstKaldiGeneric(hclg_fst_rxfilename_);
        hcl_fst_ = NULL;
        g_fst_ = NULL;
    } else {
+        KALDI_LOG << "Loading HCL and G from " << hcl_fst_rxfilename_ << " " << g_fst_rxfilename_;
        hclg_fst_ = NULL;
        hcl_fst_ = fst::StdFst::Read(hcl_fst_rxfilename_);
        g_fst_ = fst::StdFst::Read(g_fst_rxfilename_);
-
        ReadIntegerVectorSimple(disambig_rxfilename_, &disambig_);
    }

@@ -149,6 +270,7 @@ Model::Model(const char *model_path) {
        word_syms_ = g_fst_->OutputSymbols();
    }
    if (!word_syms_) {
+        KALDI_LOG << "Loading words from " << word_syms_rxfilename_;
        if (!(word_syms_ = fst::SymbolTable::ReadText(word_syms_rxfilename_)))
            KALDI_ERR << "Could not read symbol table from file "
                      << word_syms_rxfilename_;
@@ -156,11 +278,38 @@ Model::Model(const char *model_path) {
    KALDI_ASSERT(word_syms_);

    if (stat(winfo_rxfilename_.c_str(), &buffer) == 0) {
+        KALDI_LOG << "Loading winfo " << winfo_rxfilename_;
        kaldi::WordBoundaryInfoNewOpts opts;
        winfo_ = new kaldi::WordBoundaryInfo(opts, winfo_rxfilename_);
    } else {
        winfo_ = NULL;
    }
+
+    if (stat(carpa_rxfilename_.c_str(), &buffer) == 0) {
+        KALDI_LOG << "Loading CARPA model from " << carpa_rxfilename_;
+        std_lm_fst_ = fst::ReadFstKaldi(std_fst_rxfilename_);
+        fst::Project(std_lm_fst_, fst::PROJECT_OUTPUT);
+        if (std_lm_fst_->Properties(fst::kILabelSorted, true) == 0) {
+            fst::ILabelCompare<fst::StdArc> ilabel_comp;
+            fst::ArcSort(std_lm_fst_, ilabel_comp);
+        }
+        ReadKaldiObject(carpa_rxfilename_, &const_arpa_);
+    } else {
+        std_lm_fst_ = NULL;
+    }
+}
+
+void Model::Ref() 
+{
+    ref_cnt_++;
+}
+
+void Model::Unref() 
+{
+    ref_cnt_--;
+    if (ref_cnt_ == 0) {
+        delete this;
+    }
 }

 Model::~Model() {
@@ -32,6 +32,7 @@
 #include "rnnlm/rnnlm-utils.h"

 using namespace kaldi;
+using namespace std;

 class KaldiRecognizer;

@@ -39,18 +40,27 @@ class Model {

 public:
    Model(const char *model_path);
-    ~Model();
+    void Ref();
+    void Unref();

 protected:
+    ~Model();
+    void ConfigureV1();
+    void ConfigureV2();
+    void ReadDataFiles();
+
    friend class KaldiRecognizer;

-    std::string nnet3_rxfilename_;
-    std::string hclg_fst_rxfilename_;
-    std::string hcl_fst_rxfilename_;
-    std::string g_fst_rxfilename_;
-    std::string word_syms_rxfilename_;
-    std::string winfo_rxfilename_;
-    std::string disambig_rxfilename_;
+    string model_path_str_;
+    string nnet3_rxfilename_;
+    string hclg_fst_rxfilename_;
+    string hcl_fst_rxfilename_;
+    string g_fst_rxfilename_;
+    string disambig_rxfilename_;
+    string word_syms_rxfilename_;
+    string winfo_rxfilename_;
+    string carpa_rxfilename_;
+    string std_fst_rxfilename_;

    kaldi::OnlineEndpointConfig endpoint_config_;
    kaldi::LatticeFasterDecoderConfig nnet3_decoding_config_;
@@ -62,11 +72,16 @@ protected:
    kaldi::nnet3::AmNnetSimple *nnet_;
    const fst::SymbolTable *word_syms_;
    kaldi::WordBoundaryInfo *winfo_;
-    std::vector<int32> disambig_;
+    vector<int32> disambig_;

    fst::Fst<fst::StdArc> *hclg_fst_;
    fst::Fst<fst::StdArc> *hcl_fst_;
    fst::Fst<fst::StdArc> *g_fst_;
+
+    fst::VectorFst<fst::StdArc> *std_lm_fst_;
+    kaldi::ConstArpaLm const_arpa_;
+
+    int ref_cnt_;
 };

 #endif /* MODEL_H_ */
@@ -24,4 +24,19 @@ SpkModel::SpkModel(const char *speaker_path) {
    SetBatchnormTestMode(true, &speaker_nnet);
    SetDropoutTestMode(true, &speaker_nnet);
    CollapseModel(nnet3::CollapseModelConfig(), &speaker_nnet);
+
+    ref_cnt_ = 1;
+}
+
+void SpkModel::Ref() 
+{
+    ref_cnt_++;
+}
+
+void SpkModel::Unref() 
+{
+    ref_cnt_--;
+    if (ref_cnt_ == 0) {
+        delete this;
+    }
 }
@@ -27,12 +27,17 @@ class SpkModel {

 public:
    SpkModel(const char *spk_path);
+    void Ref();
+    void Unref();

 protected:
    friend class KaldiRecognizer;
+    ~SpkModel() {};

    kaldi::nnet3::Nnet speaker_nnet;
    MfccOptions spkvector_mfcc_opts;
+
+    int ref_cnt_;
 };

 #endif /* SPK_MODEL_H_ */
@@ -1,7 +1,10 @@
-%module(package="vosk") vosk
+#if SWIGPYTHON
+%module(package="vosk", "threads"=1) vosk
+#else
+%module Vosk
+#endif

 %include <typemaps.i>
-%include <std_string.i>

 #if SWIGPYTHON
 %include <pybuffer.i>
@@ -11,39 +14,12 @@
 %include <arrays_csharp.i>
 #endif

-namespace kaldi {
-}
-
 #if SWIGPYTHON
 %pybuffer_binary(const char *data, int len);
-%ignore KaldiRecognizer::AcceptWaveform(const short *sdata, int len);
-%ignore KaldiRecognizer::AcceptWaveform(const float *fdata, int len);
-%exception {
-  try {
-    $action
-  } catch (kaldi::KaldiFatalError &e) {
-    PyErr_SetString(PyExc_RuntimeError, const_cast<char*>(e.KaldiMessage()));
-    SWIG_fail;
-  } catch (std::exception &e) {
-    PyErr_SetString(PyExc_RuntimeError, const_cast<char*>(e.what()));
-    SWIG_fail;
-  }
-}
 #endif

 #if SWIGJAVA
 %apply char *BYTE {const char *data};
-%ignore KaldiRecognizer::AcceptWaveform(const short *sdata, int len);
-%ignore KaldiRecognizer::AcceptWaveform(const float *fdata, int len);
-#endif
-
-%{
-#include "kaldi_recognizer.h"
-#include "model.h"
-#include "spk_model.h"
-%}
-
-#if SWIGJAVA
 %typemap(javaimports) KaldiRecognizer %{
 import java.nio.ByteBuffer;
 import java.nio.ByteOrder;
@@ -67,6 +43,80 @@ CSHARP_ARRAYS(char, byte)
 %apply short INPUT[] {const short *sdata};
 #endif

-%include "kaldi_recognizer.h"
-%include "model.h"
-%include "spk_model.h"
+
+%{
+#include "vosk_api.h"
+typedef struct VoskModel Model;
+typedef struct VoskSpkModel SpkModel;
+typedef struct VoskRecognizer KaldiRecognizer;
+%}
+
+typedef struct {} Model;
+typedef struct {} SpkModel;
+typedef struct {} KaldiRecognizer;
+
+%extend Model {
+    Model(const char *model_path)  {
+        return vosk_model_new(model_path);
+    }
+    ~Model() {
+        vosk_model_free($self);
+    }
+}
+
+%extend SpkModel {
+    SpkModel(const char *model_path)  {
+        return vosk_spk_model_new(model_path);
+    }
+    ~SpkModel() {
+        vosk_spk_model_free($self);
+    }
+}
+
+%extend KaldiRecognizer {
+    KaldiRecognizer(Model *model, float sample_rate)  {
+        return vosk_recognizer_new(model, sample_rate);
+    }
+    KaldiRecognizer(Model *model, SpkModel *spk_model, float sample_rate)  {
+        return vosk_recognizer_new_spk(model, spk_model, sample_rate);
+    }
+    KaldiRecognizer(Model *model, float sample_rate, const char* grammar)  {
+        return vosk_recognizer_new_grm(model, sample_rate, grammar);
+    }
+    ~KaldiRecognizer() {
+        vosk_recognizer_free($self);
+    }
+
+#if SWIGCSHARP
+    bool AcceptWaveform(const char *data, int len) {
+        return vosk_recognizer_accept_waveform($self, data, len);
+    }
+    bool AcceptWaveform(const short *sdata, int len) {
+        return vosk_recognizer_accept_waveform_s($self, sdata, len);
+    }
+    bool AcceptWaveform(const float *fdata, int len) {
+        return vosk_recognizer_accept_waveform_f($self, fdata, len);
+    }
+#elif SWIGJAVA
+    bool AcceptWaveform(const char *data, int len) {
+        return vosk_recognizer_accept_waveform($self, data, len);
+    }
+#else
+    int AcceptWaveform(const char *data, int len) {
+        return vosk_recognizer_accept_waveform($self, data, len);
+    }
+#endif
+
+    const char* Result() {
+        return vosk_recognizer_result($self);
+    }
+    const char* PartialResult() {
+        return vosk_recognizer_partial_result($self);
+    }
+    const char* FinalResult() {
+        return vosk_recognizer_final_result($self);
+    }
+}
+
+%rename(SetLogLevel) vosk_set_log_level;
+void vosk_set_log_level(int level);
@@ -0,0 +1,97 @@
+// Copyright 2020 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.
+
+#include "vosk_api.h"
+#include "kaldi_recognizer.h"
+#include "model.h"
+#include "spk_model.h"
+
+#include <string.h>
+
+using namespace kaldi;
+
+VoskModel *vosk_model_new(const char *model_path)
+{
+    return (VoskModel *)new Model(model_path);
+}
+
+void vosk_model_free(VoskModel *model)
+{
+    ((Model *)model)->Unref();
+}
+
+VoskSpkModel *vosk_spk_model_new(const char *model_path)
+{
+    return (VoskSpkModel *)new SpkModel(model_path);
+}
+
+void vosk_spk_model_free(VoskSpkModel *model)
+{
+    ((SpkModel *)model)->Unref();
+}
+
+VoskRecognizer *vosk_recognizer_new(VoskModel *model, float sample_rate)
+{
+    return (VoskRecognizer *)new KaldiRecognizer((Model *)model, sample_rate);
+}
+
+VoskRecognizer *vosk_recognizer_new_spk(VoskModel *model, VoskSpkModel *spk_model, float sample_rate)
+{
+    return (VoskRecognizer *)new KaldiRecognizer((Model *)model, (SpkModel *)spk_model, sample_rate);
+}
+
+VoskRecognizer *vosk_recognizer_new_grm(VoskModel *model, float sample_rate, const char *grammar)
+{
+    return (VoskRecognizer *)new KaldiRecognizer((Model *)model, sample_rate, grammar);
+}
+
+int vosk_recognizer_accept_waveform(VoskRecognizer *recognizer, const char *data, int length)
+{
+    return ((KaldiRecognizer *)(recognizer))->AcceptWaveform(data, length);
+}
+
+int vosk_recognizer_accept_waveform_s(VoskRecognizer *recognizer, const short *data, int length)
+{
+    return ((KaldiRecognizer *)(recognizer))->AcceptWaveform(data, length);
+}
+
+int vosk_recognizer_accept_waveform_f(VoskRecognizer *recognizer, const float *data, int length)
+{
+    return ((KaldiRecognizer *)(recognizer))->AcceptWaveform(data, length);
+}
+
+const char *vosk_recognizer_result(VoskRecognizer *recognizer)
+{
+    return ((KaldiRecognizer *)recognizer)->Result();
+}
+
+const char *vosk_recognizer_partial_result(VoskRecognizer *recognizer)
+{
+    return ((KaldiRecognizer *)recognizer)->PartialResult();
+}
+
+const char *vosk_recognizer_final_result(VoskRecognizer *recognizer)
+{
+    return ((KaldiRecognizer *)recognizer)->FinalResult();
+}
+
+void vosk_recognizer_free(VoskRecognizer *recognizer)
+{
+    delete (KaldiRecognizer *)(recognizer);
+}
+
+void vosk_set_log_level(int log_level)
+{
+    SetVerboseLevel(log_level);
+}
@@ -0,0 +1,56 @@
+// Copyright 2020 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.
+
+
+#ifndef _VOSK_API_H_
+#define _VOSK_API_H_
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+typedef struct VoskModel VoskModel;
+typedef struct VoskSpkModel VoskSpkModel;
+typedef struct VoskRecognizer VoskRecognizer;
+
+VoskModel *vosk_model_new(const char *model_path);
+void vosk_model_free(VoskModel *model);
+
+VoskSpkModel *vosk_spk_model_new(const char *model_path);
+void vosk_spk_model_free(VoskSpkModel *model);
+
+VoskRecognizer *vosk_recognizer_new(VoskModel *model, float sample_rate);
+VoskRecognizer *vosk_recognizer_new_spk(VoskModel *model, VoskSpkModel *spk_model, float sample_rate);
+VoskRecognizer *vosk_recognizer_new_grm(VoskModel *model, float sample_rate, const char *grammar);
+int vosk_recognizer_accept_waveform(VoskRecognizer *recognizer, const char *data, int length);
+int vosk_recognizer_accept_waveform_s(VoskRecognizer *recognizer, const short *data, int length);
+int vosk_recognizer_accept_waveform_f(VoskRecognizer *recognizer, const float *data, int length);
+const char *vosk_recognizer_result(VoskRecognizer *recognizer);
+const char *vosk_recognizer_partial_result(VoskRecognizer *recognizer);
+const char *vosk_recognizer_final_result(VoskRecognizer *recognizer);
+void vosk_recognizer_free(VoskRecognizer *recognizer);
+
+/**
+ * Set log level for Kaldi messages
+ *   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_ */
@@ -21,6 +21,7 @@ RUN cd /opt \
    && rm -rf swig-4.0.1.tar.gz swig-4.0.1

 ARG OPENBLAS_ARCH=ARMV7
+ARG ARM_HARDWARE_OPTS="-mfloat-abi=hard -mfpu=neon"
 RUN cd /opt \
    && export OPENFST_CONFIGURE="--enable-static --enable-shared --enable-far --enable-ngram-fsts --enable-lookahead-fsts --with-pic --disable-bin --host=${CROSS_TRIPLE} --build=x86-linux-gnu" \
    && git clone -b lookahead --single-branch https://github.com/alphacep/kaldi \
@@ -31,8 +32,10 @@ RUN cd /opt \
    && make -j 10 openfst \
    && cd /opt/kaldi/src \
    && sed -i "s:TARGET_ARCH=\"\`uname -m\`\":TARGET_ARCH=$(echo $CROSS_TRIPLE|cut -d - -f 1):g" configure \
+    && sed -i "s:-mfloat-abi=hard -mfpu=neon:${ARM_HARDWARE_OPTS}:g" makefiles/linux_openblas_arm.mk \
+    && sed -i "s: -O1 : -O3 :g" makefiles/linux_openblas_arm.mk \
    && ./configure --mathlib=OPENBLAS --shared --use-cuda=no \
-    && make -j 10 online2 \
+    && make -j 10 online2 lm \
    && find /opt/kaldi -name "*.o" -exec rm {} \;

 RUN cd /opt \
@@ -60,3 +63,20 @@ RUN cd /opt \
    && make -j $(nproc) \
    && make install \
    && rm -rf /opt/cpython-3.7.6 /opt/cpython-3.7.6-cross /opt/v3.7.6.tar.gz
+
+RUN cd /opt \
+    && wget -q https://github.com/python/cpython/archive/v3.6.10.tar.gz \
+    && tar xf v3.6.10.tar.gz \
+    && cp -r cpython-3.6.10 cpython-3.6.10-cross \
+    && cd /opt/cpython-3.6.10 \
+    && AR=/usr/bin/ar RANLIB=/usr/bin/ranlib CPP=/usr/bin/cpp CXX=/usr/bin/g++ CC=/usr/bin/gcc  ./configure --prefix="/opt/python/cp3.6-cp3.6m" \
+    && make -j $(nproc) \
+    && make install \
+    && /opt/python/cp3.6-cp3.6m/bin/pip3 install -U pip \
+    && /opt/python/cp3.6-cp3.6m/bin/pip3 install -U wheel \
+    && cd /opt/cpython-3.6.10-cross \
+    && export PATH=/opt/python/cp3.6-cp3.6m/bin:$PATH \
+    && ./configure --prefix=$CROSS_ROOT --with-openssl=$CROSS_ROOT --host=${CROSS_TRIPLE} --build=x86-linux-gnu --disable-ipv6 ac_cv_file__dev_ptmx=no ac_cv_file__dev_ptc=no ac_cv_have_long_long_format=yes \
+    && make -j $(nproc) \
+    && make install \
+    && rm -rf /opt/cpython-3.6.10 /opt/cpython-3.6.10-cross /opt/v3.6.10.tar.gz
@@ -27,7 +27,7 @@ RUN cd /opt \
    && make -j 10 openfst \
    && cd ../src \
    && ./configure --mathlib=OPENBLAS --shared --use-cuda=no \
-    && make -j 10 online2 \
+    && make -j 10 online2 lm \
    && find /opt/kaldi -name "*.o" -exec rm {} \;

 RUN cd /opt \
@@ -37,4 +37,3 @@ RUN cd /opt \
    && ./configure --prefix=/usr && make -j 10 && make install \
    && cd .. \
    && rm -rf swig-4.0.1.tar.gz swig-4.0.1
-
@@ -4,5 +4,9 @@ set -e
 set -x

 docker build --build-arg="DOCKCROSS_IMAGE=linux-armv7" --build-arg="OPENBLAS_ARCH=ARMV7" --file Dockerfile.dockcross --tag alphacep/kaldi-dockcross-armv7:latest .
-docker build --build-arg="DOCKCROSS_IMAGE=linux-armv6" --build-arg="OPENBLAS_ARCH=ARMV6" --file Dockerfile.dockcross --tag alphacep/kaldi-dockcross-armv6:latest .
+docker build --build-arg="DOCKCROSS_IMAGE=linux-armv6" --build-arg="OPENBLAS_ARCH=ARMV6" --build-arg="ARM_HARDWARE_OPTS=" --file Dockerfile.dockcross --tag alphacep/kaldi-dockcross-armv6:latest .
 docker build --build-arg="DOCKCROSS_IMAGE=linux-arm64" --build-arg="OPENBLAS_ARCH=ARMV8" --file Dockerfile.dockcross --tag alphacep/kaldi-dockcross-arm64:latest .
+
+docker run --rm -v /home/shmyrev/travis/vosk-api/:/io alphacep/kaldi-dockcross-armv6 /io/travis/build-wheels-dockcross.sh
+docker run --rm -v /home/shmyrev/travis/vosk-api/:/io alphacep/kaldi-dockcross-armv7 /io/travis/build-wheels-dockcross.sh
+docker run --rm -v /home/shmyrev/travis/vosk-api/:/io alphacep/kaldi-dockcross-arm64 /io/travis/build-wheels-dockcross.sh
@@ -4,3 +4,4 @@ set -e
 set -x

 docker build --file Dockerfile.manylinux --tag alphacep/kaldi-manylinux:latest .
+docker run --rm -e PLAT=manylinux2010_x86_64 -v /home/shmyrev/travis/vosk-api/:/io alphacep/kaldi-manylinux /io/travis/build-wheels.sh
@@ -1,20 +1,25 @@
 #!/bin/bash
 set -e -x

-export KALDI_ROOT=/opt/kaldi
-export WHEEL_FLAGS=`$CROSS_ROOT/bin/python3-config --cflags`
-export PATH=/opt/python/cp3.7-cp3.7m/bin:$PATH
-echo $CROSS_TRIPLE
-case $CROSS_TRIPLE in
-    *arm-*)
-        export _PYTHON_HOST_PLATFORM=linux-armv6l
-        ;;
-    *armv7-*)
-        export _PYTHON_HOST_PLATFORM=linux-armv7l
-        ;;
-    *aarch64-*)
-        export _PYTHON_HOST_PLATFORM=linux-aarch64
-        ;;
-esac
+ORIG_PATH=$PATH
+for pyver in 3.6 3.7; do

-pip3 wheel /io/python -w /io/wheelhouse
+    export KALDI_ROOT=/opt/kaldi
+    export WHEEL_FLAGS=`$CROSS_ROOT/bin/python${pyver}-config --cflags`
+    export PATH=/opt/python/cp${pyver}-cp${pyver}m/bin:$ORIG_PATH
+    echo $CROSS_TRIPLE
+    case $CROSS_TRIPLE in
+        *arm-*)
+            export _PYTHON_HOST_PLATFORM=linux-armv6l
+            ;;
+        *armv7-*)
+            export _PYTHON_HOST_PLATFORM=linux-armv7l
+            ;;
+        *aarch64-*)
+            export _PYTHON_HOST_PLATFORM=linux-aarch64
+            ;;
+    esac
+
+    pip3 wheel /io/python -w /io/wheelhouse
+
+done
@@ -4,7 +4,7 @@ set -e -x
 export KALDI_ROOT=/opt/kaldi

 # Compile wheels
-for pypath in /opt/python/cp3*; do
+for pypath in /opt/python/cp3[56789]*; do
    export WHEEL_FLAGS=`${pypath}/bin/python3-config --cflags`
    mkdir -p /opt/wheelhouse
    "${pypath}/bin/pip" wheel /io/python -w /opt/wheelhouse
Author	SHA1	Message	Date
Nickolay Shmyrev	5c4dd4644e	Update version	2020-05-03 20:52:43 +02:00
Nickolay Shmyrev	9bbd172cfd	Some more links	2020-05-02 01:37:10 +02:00
Nickolay Shmyrev	dbf9de77c3	Document model training and model structure	2020-05-02 01:11:25 +02:00
Nickolay Shmyrev	8b790cd162	Better android logging	2020-05-01 20:02:46 +02:00
Nickolay Shmyrev	80219066e9	Expose verbose level in the API	2020-05-01 19:02:57 +02:00
Nickolay Shmyrev	26fa5f098f	Updated link to English model	2020-04-30 15:19:13 +02:00
Nickolay Shmyrev	d75bb36131	Fix model download path	2020-04-30 11:04:35 +02:00
Nickolay Shmyrev	30c5e8ca79	Link to lm library for rescoring	2020-04-30 10:52:15 +02:00
Nickolay Shmyrev	c00e36fab6	Rearrange models	2020-04-30 10:47:10 +02:00
Nickolay Shmyrev	80e60b9118	LM Rescoring	2020-04-28 22:17:34 +02:00
Nickolay Shmyrev	e3a95a44bc	Added Turkish model	2020-04-28 10:10:35 +02:00
Nickolay Shmyrev	86caf526f5	New model layout and model config file with beams for decoding	2020-04-28 01:06:41 +02:00
Nickolay Shmyrev	e09f32b4b4	Properly stop listening so it can continue later	2020-04-27 09:00:32 +02:00
Nickolay Shmyrev	3803ab345d	Proper reference couting of the models to avoid memory issus	2020-04-23 23:37:52 +02:00
Nickolay V. Shmyrev	889b43136f	Update README.zh.md	2020-04-23 12:43:00 +02:00
John Baber-Lucero	b517cf46af	Fix some grammar/punctuation	2020-04-23 00:28:34 -04:00
Nickolay Shmyrev	ffd810fe00	Fix threading bug	2020-04-22 22:15:45 +02:00
Nickolay V. Shmyrev	a1a0ed70a1	Update README.zh.md	2020-04-22 18:29:01 +03:00
Nickolay Shmyrev	1554d9ede7	Added per-language readme	2020-04-22 13:21:36 +02:00
Nickolay Shmyrev	6f3190d83d	Version 0.3.4 and arm wheels for python 3.6	2020-04-21 11:38:56 +02:00
Nickolay Shmyrev	3caaa32ec0	How to ask for the accuracy updates	2020-04-20 23:11:52 +02:00
Nickolay Shmyrev	3facf3ccf5	UI updates	2020-04-20 11:08:18 +03:00
Nickolay Shmyrev	fad954e6e3	Full iOS project	2020-04-19 22:48:47 +03:00
Nickolay Shmyrev	c4809cb618	Added Vietnamese model	2020-04-19 20:14:05 +02:00
Nickolay Shmyrev	a241423baf	Added iOS bits	2020-04-18 17:34:03 +03:00
Nickolay Shmyrev	1e9421dd38	Add a note about iOS	2020-04-18 15:41:51 +02:00
Nickolay Shmyrev	b09ffda760	Fixes links and issue #59	2020-04-11 22:15:46 +02:00
Nickolay Shmyrev	aeff663a7c	Fix pocketsphinx references in Android sources	2020-04-09 21:50:38 +02:00
Nickolay Shmyrev	fcd17fcd4c	Added Russian model 0.9	2020-04-07 17:48:29 +02:00
Nickolay Shmyrev	feffb2711d	Repair travis build	2020-04-04 20:49:06 +02:00
Nickolay Shmyrev	f76e5b592f	kaldi-ru-0.8	2020-04-04 10:27:02 +02:00
Nickolay Shmyrev	71bdc900e3	C-only wrapper	2020-04-04 10:22:01 +02:00
Nickolay Shmyrev	96bbf5abc2	Add C API	2020-04-04 10:04:49 +02:00
Nickolay Shmyrev	df9424a228	Introduce pure C api to deal with Windows runtime issues and make it easy to wrap in Swift	2020-04-04 00:35:56 +02:00
Nickolay Shmyrev	3310acaf54	Add lm to kaldi-android	2020-03-28 22:45:38 +01:00
Nickolay V. Shmyrev	e8722d462d	Merge pull request #54 from andremendesc/master Fixes folder name instruction for python model on microphone test	2020-03-25 08:37:25 +03:00
André Mendes	14b2c13ed6	Fixes folder name instruction for python model on microphone test	2020-03-25 01:01:23 -03:00
Nickolay Shmyrev	19af324096	Small fix by Funny Jingl	2020-03-19 23:24:41 +01:00
Nickolay Shmyrev	0ca7b94e08	Add link on kaldi-ru-0.7	2020-03-19 20:14:41 +01:00
Nickolay Shmyrev	04ed310229	Move models list to doc	2020-03-15 22:39:57 +01:00
Nickolay Shmyrev	7ac33d521c	Added information about models	2020-03-15 22:27:47 +01:00
Nickolay V. Shmyrev	08ada63da4	Added note about other projects	2020-03-15 02:48:49 +03:00
Nickolay Shmyrev	cef3fd72fb	Disable NEON fpu on ARMv6 Fixes issue #46	2020-03-10 00:24:57 +01:00
Nickolay V. Shmyrev	bf973ff434	Merge pull request #43 from camillem/patch-1 Typo in folder name	2020-03-02 15:47:25 +03:00
camillem	a172d60b20	Typo in folder name	2020-03-02 13:46:34 +01:00
Nickolay Shmyrev	060e4395c2	Fix repeating results issue #42 Thanks to Yondu Tsai	2020-03-02 13:41:27 +01:00
Nickolay Shmyrev	03f1417454	Added small readme	2020-03-01 09:12:09 +01:00