Compare commits
25 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
| a8ae6025bd | |||
| 1b8332a609 | |||
| fe16ec7e57 | |||
| c5ce5e46bd | |||
| 99d66670bc | |||
| 072b42cac6 | |||
| acadb5b4c2 | |||
| 2e1de6c3af | |||
| 67de30908b | |||
| 9d398eff0e | |||
| 02b7312a00 | |||
| 08c35e84f3 | |||
| b92a5c1fc7 | |||
| 8997a587c5 | |||
| dc3d03d742 | |||
| 84df40715c | |||
| b639fb501a | |||
| 746ff47757 | |||
| 2d62db8118 | |||
| 7a2adcd9ba | |||
| 9ccf3ef0e8 | |||
| 0edab6d558 | |||
| 4b344f0cd8 | |||
| 17ddc4d5ba | |||
| 3e33860c47 |
@@ -53,6 +53,7 @@ csharp/*.exe
|
||||
csharp/*.c
|
||||
csharp/model/
|
||||
csharp/test.wav
|
||||
Vosk.dll
|
||||
|
||||
# Node
|
||||
nodejs/vosk_wrap.cc
|
||||
|
||||
@@ -1,9 +1,9 @@
|
||||
# About
|
||||
|
||||
Vosk is an offline open source speech recognition toolkit. It enables
|
||||
speech recognition models for 16 languages and dialects - English, Indian
|
||||
speech recognition models for 17 languages and dialects - English, Indian
|
||||
English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish,
|
||||
Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi.
|
||||
Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi, Filipino.
|
||||
|
||||
Vosk models are small (50 Mb) but provide continuous large vocabulary
|
||||
transcription, zero-latency response with streaming API, reconfigurable
|
||||
|
||||
@@ -32,7 +32,7 @@ set(API_SOURCES
|
||||
"${PROJECT_SOURCE_DIR}/../src/vosk_api.h"
|
||||
)
|
||||
|
||||
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O3 -DFST_NO_DYNAMIC_LINKING")
|
||||
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++17 -O3 -DFST_NO_DYNAMIC_LINKING")
|
||||
|
||||
add_library( vosk_jni SHARED
|
||||
build/generated-src/cpp/vosk_wrap.cc
|
||||
@@ -60,7 +60,7 @@ target_link_libraries( vosk_jni
|
||||
${KALDI_ROOT}/src/base/kaldi-base.a
|
||||
${LIB_ROOT}/lib/libfst.a
|
||||
${LIB_ROOT}/lib/libfstngram.a
|
||||
${LIB_ROOT}/lib/libopenblas_${OPENBLAS_ARCH}-r0.3.7.a
|
||||
${LIB_ROOT}/lib/libopenblas.a
|
||||
${LIB_ROOT}/lib/libclapack.a
|
||||
${LIB_ROOT}/lib/liblapack.a
|
||||
${LIB_ROOT}/lib/libblas.a
|
||||
|
||||
+2
-21
@@ -1,22 +1,3 @@
|
||||
This is still work in progress, more to come
|
||||
Vosk library for Android
|
||||
|
||||
## TODO
|
||||
|
||||
* Optimize graph construction, current one is below accuracy
|
||||
|
||||
* Load model from the AAR (mmap them in tflite style)
|
||||
|
||||
* Add decoding speed measurement
|
||||
|
||||
* Add wakeup word
|
||||
|
||||
* Add speakerid
|
||||
|
||||
* Integrate proper hardware optimized neural network library. Candidates are:
|
||||
|
||||
* https://github.com/XiaoMi/mace
|
||||
* https://github.com/Tencent/ncnn
|
||||
* https://developer.android.com/ndk/guides/neuralnetworks/ (since API level 27)
|
||||
* https://github.com/google/XNNPACK
|
||||
|
||||
* Quantization for the models
|
||||
See for details https://alphacephei.com/vosk/android
|
||||
|
||||
@@ -39,7 +39,7 @@ WORKDIR_X86_64=`pwd`/build/kaldi_x86_64
|
||||
WORKDIR_ARM32=`pwd`/build/kaldi_arm_32
|
||||
WORKDIR_ARM64=`pwd`/build/kaldi_arm_64
|
||||
PATH=$PATH:$ANDROID_NDK_HOME/toolchains/llvm/prebuilt/${OS_NAME}-x86_64/bin
|
||||
OPENFST_VERSION=1.6.7
|
||||
OPENFST_VERSION=1.8.0
|
||||
|
||||
mkdir -p $WORKDIR_ARM64/local/lib $WORKDIR_ARM32/local/lib $WORKDIR_X86_64/local/lib $WORKDIR_X86/local/lib
|
||||
|
||||
@@ -106,7 +106,7 @@ esac
|
||||
|
||||
# openblas first
|
||||
cd $WORKDIR
|
||||
git clone -b v0.3.7 --single-branch https://github.com/xianyi/OpenBLAS
|
||||
git clone -b v0.3.13 --single-branch https://github.com/xianyi/OpenBLAS
|
||||
make -C OpenBLAS TARGET=$BLAS_ARCH ONLY_CBLAS=1 AR=$AR CC=$CC HOSTCC=gcc ARM_SOFTFP_ABI=1 USE_THREAD=0 NUM_THREADS=1 -j4
|
||||
make -C OpenBLAS install PREFIX=$WORKDIR/local
|
||||
|
||||
@@ -130,7 +130,7 @@ if [ "`uname`" == "Darwin" ]; then
|
||||
fi
|
||||
|
||||
CXX=$CXX CXXFLAGS="$ARCHFLAGS -O3 -DFST_NO_DYNAMIC_LINKING" ./configure --use-cuda=no \
|
||||
--mathlib=OPENBLAS --shared \
|
||||
--mathlib=OPENBLAS_CLAPACK --shared \
|
||||
--android-incdir=${ANDROID_TOOLCHAIN_PATH}/sysroot/usr/include \
|
||||
--host=$HOST --openblas-root=${WORKDIR}/local \
|
||||
--fst-root=${WORKDIR}/local --fst-version=${OPENFST_VERSION}
|
||||
|
||||
@@ -14,7 +14,7 @@ plugins {
|
||||
}
|
||||
|
||||
def archiveName = "vosk-android"
|
||||
def libVersion = "0.3.15"
|
||||
def libVersion = "0.3.17"
|
||||
|
||||
allprojects {
|
||||
repositories {
|
||||
|
||||
@@ -0,0 +1,71 @@
|
||||
buildscript {
|
||||
repositories {
|
||||
google()
|
||||
jcenter()
|
||||
}
|
||||
dependencies {
|
||||
classpath 'com.android.tools.build:gradle:3.5.3'
|
||||
classpath 'com.jfrog.bintray.gradle:gradle-bintray-plugin:1.8.5'
|
||||
}
|
||||
}
|
||||
|
||||
plugins {
|
||||
id "com.jfrog.bintray"
|
||||
}
|
||||
|
||||
def archiveName = "vosk-model-en"
|
||||
def libVersion = "0.3.17"
|
||||
|
||||
allprojects {
|
||||
repositories {
|
||||
google()
|
||||
jcenter()
|
||||
}
|
||||
}
|
||||
|
||||
apply plugin: 'com.android.library'
|
||||
apply plugin: 'maven-publish'
|
||||
|
||||
android {
|
||||
compileSdkVersion 29
|
||||
defaultConfig {
|
||||
minSdkVersion 21
|
||||
targetSdkVersion 29
|
||||
versionCode 6
|
||||
versionName = libVersion
|
||||
archivesBaseName = archiveName
|
||||
}
|
||||
}
|
||||
|
||||
Properties properties = new Properties()
|
||||
properties.load(project.rootProject.file('local.properties').newDataInputStream())
|
||||
|
||||
bintray {
|
||||
user = properties.getProperty("bintray.user")
|
||||
key = properties.getProperty("bintray.apikey")
|
||||
pkg {
|
||||
repo = 'vosk'
|
||||
name = 'vosk-model-en'
|
||||
userOrg = "alphacep"
|
||||
licenses = ['Apache2.0']
|
||||
websiteUrl = 'https://github.com/alphacep/vosk-api'
|
||||
issueTrackerUrl = 'https://github.com/alphacep/vosk-api/issues'
|
||||
vcsUrl = 'https://github.com/alphacep/vosk-api'
|
||||
version {
|
||||
name = libVersion
|
||||
vcsTag = libVersion
|
||||
}
|
||||
}
|
||||
publications = ['aar']
|
||||
}
|
||||
|
||||
publishing {
|
||||
publications {
|
||||
aar(MavenPublication) {
|
||||
groupId 'com.alphacep'
|
||||
artifactId archiveName
|
||||
version libVersion
|
||||
artifact("$buildDir/outputs/aar/$archiveName-release.aar")
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,3 @@
|
||||
<manifest xmlns:android="http://schemas.android.com/apk/res/android"
|
||||
package="org.kaldi.model.en">
|
||||
</manifest>
|
||||
@@ -0,0 +1,7 @@
|
||||
US English model for mobile Vosk applications
|
||||
|
||||
Copyright 2020 Alpha Cephei Inc
|
||||
|
||||
Accuracy: 10.38 (tedlium test) 9.85 (librispeech test-clean)
|
||||
Speed: 0.11xRT (desktop)
|
||||
Latency: 0.15s (right context)
|
||||
@@ -0,0 +1 @@
|
||||
0.3.16
|
||||
@@ -0,0 +1 @@
|
||||
include 'model-en'
|
||||
+8
-5
@@ -25,17 +25,20 @@ LIBS= \
|
||||
$(KALDI_ROOT)/src/fstext/kaldi-fstext.a \
|
||||
$(KALDI_ROOT)/src/util/kaldi-util.a \
|
||||
$(KALDI_ROOT)/src/base/kaldi-base.a \
|
||||
$(KALDI_ROOT)/tools/OpenBLAS/libopenblas.a \
|
||||
$(KALDI_ROOT)/tools/OpenBLAS/install/lib/libopenblas.a \
|
||||
$(KALDI_ROOT)/tools/OpenBLAS/install/lib/liblapack.a \
|
||||
$(KALDI_ROOT)/tools/OpenBLAS/install/lib/libblas.a \
|
||||
$(KALDI_ROOT)/tools/OpenBLAS/install/lib/libf2c.a \
|
||||
$(KALDI_ROOT)/tools/openfst/lib/libfst.a \
|
||||
$(KALDI_ROOT)/tools/openfst/lib/libfstngram.a
|
||||
|
||||
all: test_vosk test_vosk_speaker
|
||||
|
||||
test_vosk: test_vosk.o libvosk.a
|
||||
g++ $^ -o $@ $(LIBS) -lgfortran -lpthread
|
||||
g++ $^ -o $@ $(LIBS) -lpthread
|
||||
|
||||
test_vosk_speaker: test_vosk_speaker.o libvosk.a
|
||||
g++ $^ -o $@ $(LIBS) -lgfortran -lpthread
|
||||
g++ $^ -o $@ $(LIBS) -lpthread
|
||||
|
||||
libvosk.a: $(VOSK_SOURCES:.cc=.o)
|
||||
ar rcs $@ $^
|
||||
@@ -44,7 +47,7 @@ libvosk.a: $(VOSK_SOURCES:.cc=.o)
|
||||
g++ $(CFLAGS) -c -o $@ $<
|
||||
|
||||
%.o: %.cc
|
||||
g++ -std=c++11 $(CFLAGS) -c -o $@ $<
|
||||
g++ -std=c++17 $(CFLAGS) -c -o $@ $<
|
||||
|
||||
clean:
|
||||
rm -f *.o *.a test_vosk
|
||||
rm -f *.o *.a test_vosk test_vosk_speaker
|
||||
|
||||
@@ -24,5 +24,6 @@ int main() {
|
||||
|
||||
vosk_recognizer_free(recognizer);
|
||||
vosk_model_free(model);
|
||||
fclose(wavin);
|
||||
return 0;
|
||||
}
|
||||
|
||||
+10
-4
@@ -1,5 +1,5 @@
|
||||
KALDI_ROOT ?= $(HOME)/kaldi
|
||||
CFLAGS := -std=c++11 -g -O2 -DPIC -fPIC -Wno-unused-function
|
||||
CFLAGS := -std=c++17 -g -O2 -DPIC -fPIC -Wno-unused-function
|
||||
CPPFLAGS := -I$(KALDI_ROOT)/src -I$(KALDI_ROOT)/tools/openfst/include -I../src -DFST_NO_DYNAMIC_LINKING
|
||||
|
||||
KALDI_LIBS = \
|
||||
@@ -23,14 +23,20 @@ KALDI_LIBS = \
|
||||
${KALDI_ROOT}/tools/openfst/lib/libfstngram.a
|
||||
|
||||
# On Linux
|
||||
MATH_LIBS = ${KALDI_ROOT}/tools/OpenBLAS/libopenblas.a -lgfortran
|
||||
MATH_LIBS = \
|
||||
${KALDI_ROOT}/tools/OpenBLAS/install/lib/libopenblas.a \
|
||||
${KALDI_ROOT}/tools/OpenBLAS/install/lib/liblapack.a \
|
||||
${KALDI_ROOT}/tools/OpenBLAS/install/lib/libblas.a \
|
||||
${KALDI_ROOT}/tools/OpenBLAS/install/lib/libf2c.a
|
||||
|
||||
# On OSX
|
||||
# MATH_LIBS = -framework Accelerate
|
||||
|
||||
all: test.exe
|
||||
|
||||
test.exe: libkaldiwrap.so test.cs
|
||||
mcs test.cs gen/*.cs
|
||||
mcs -out:Vosk.dll -target:library gen/*.cs
|
||||
mcs test.cs -r:Vosk.dll -out:test.exe
|
||||
|
||||
VOSK_SOURCES = \
|
||||
vosk_wrap.c \
|
||||
@@ -59,4 +65,4 @@ run: test.exe
|
||||
mono test.exe
|
||||
|
||||
clean:
|
||||
$(RM) *.so vosk_wrap.c *.o gen/*.cs test.exe
|
||||
$(RM) *.so vosk_wrap.c *.o gen/*.cs test.exe *.dll
|
||||
|
||||
@@ -0,0 +1,27 @@
|
||||
<?xml version="1.0"?>
|
||||
<package>
|
||||
<metadata>
|
||||
<id>Vosk</id>
|
||||
<version>0.3.16</version>
|
||||
<authors>Alpha Cephei Inc</authors>
|
||||
<owners>Alpha Cephei Inc</owners>
|
||||
<license type="expression">Apache-2.0</license>
|
||||
<projectUrl>https://alphacephei.com/vosk/</projectUrl>
|
||||
<requireLicenseAcceptance>false</requireLicenseAcceptance>
|
||||
<description>Vosk is an offline open source speech recognition toolkit. It enables speech recognition models for 16 languages and dialects - English, Indian English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish, Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi.
|
||||
|
||||
Vosk models are small (50 Mb) but provide continuous large vocabulary transcription, zero-latency response with streaming API, reconfigurable vocabulary and speaker identification.
|
||||
|
||||
Speech recognition bindings implemented for various programming languages like Python, Java, Node.JS, C#, C++ and others.
|
||||
|
||||
Vosk supplies speech recognition for chatbots, smart home appliances, virtual assistants. It can also create subtitles for movies, transcription for lectures and interviews.
|
||||
|
||||
Vosk scales from small devices like Raspberry Pi or Android smartphone to big clusters.</description>
|
||||
<releaseNotes>See for details https://github.com/alphacep/vosk-api/releases</releaseNotes>
|
||||
<copyright>Copyright 2020 Alpha Cephei Inc</copyright>
|
||||
<tags>speech recognition voice stt asr speech-to-text ai offline privacy</tags>
|
||||
<dependencies>
|
||||
<group targetFramework=".NETStandard2.0"/>
|
||||
</dependencies>
|
||||
</metadata>
|
||||
</package>
|
||||
@@ -0,0 +1,9 @@
|
||||
<Project xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
|
||||
<ItemGroup>
|
||||
<Content Include="$(MSBuildThisFileDirectory)lib/**/**">
|
||||
<Link>%(RecursiveDir)%(Filename)%(Extension)</Link>
|
||||
<CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
|
||||
<Visible>false</Visible>
|
||||
</Content>
|
||||
</ItemGroup>
|
||||
</Project>
|
||||
@@ -16,56 +16,11 @@
|
||||
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 */; };
|
||||
92D86BD5253F823F0040D53F /* vosk-model-small-en-us-0.4 in Resources */ = {isa = PBXBuildFile; fileRef = 92D86BD3253F823E0040D53F /* vosk-model-small-en-us-0.4 */; };
|
||||
92D86BD6253F823F0040D53F /* vosk-model-spk-0.4 in Resources */ = {isa = PBXBuildFile; fileRef = 92D86BD4253F823F0040D53F /* vosk-model-spk-0.4 */; };
|
||||
/* End PBXBuildFile section */
|
||||
|
||||
/* Begin PBXCopyFilesBuildPhase section */
|
||||
92375265240C6ECF00DD6076 /* CopyFiles */ = {
|
||||
isa = PBXCopyFilesBuildPhase;
|
||||
buildActionMask = 2147483647;
|
||||
dstPath = "model-en";
|
||||
dstSubfolderSpec = 7;
|
||||
files = (
|
||||
92375266240C6EFE00DD6076 /* disambig_tid.int in CopyFiles */,
|
||||
92375267240C6EFE00DD6076 /* final.mdl in CopyFiles */,
|
||||
92375268240C6EFE00DD6076 /* Gr.fst in CopyFiles */,
|
||||
92375269240C6EFE00DD6076 /* HCLr.fst in CopyFiles */,
|
||||
9237526A240C6EFE00DD6076 /* mfcc.conf in CopyFiles */,
|
||||
9237526B240C6EFE00DD6076 /* word_boundary.int in CopyFiles */,
|
||||
9237526C240C6EFE00DD6076 /* words.txt in CopyFiles */,
|
||||
);
|
||||
runOnlyForDeploymentPostprocessing = 0;
|
||||
};
|
||||
9237526D240C6F0400DD6076 /* CopyFiles */ = {
|
||||
isa = PBXCopyFilesBuildPhase;
|
||||
buildActionMask = 2147483647;
|
||||
dstPath = "model-en/ivector";
|
||||
dstSubfolderSpec = 7;
|
||||
files = (
|
||||
9237526E240C6F1500DD6076 /* final.dubm in CopyFiles */,
|
||||
9237526F240C6F1500DD6076 /* final.ie in CopyFiles */,
|
||||
92375270240C6F1500DD6076 /* final.mat in CopyFiles */,
|
||||
92375271240C6F1500DD6076 /* global_cmvn.stats in CopyFiles */,
|
||||
92375272240C6F1500DD6076 /* online_cmvn.conf in CopyFiles */,
|
||||
92375273240C6F1500DD6076 /* splice.conf in CopyFiles */,
|
||||
);
|
||||
runOnlyForDeploymentPostprocessing = 0;
|
||||
};
|
||||
/* End PBXCopyFilesBuildPhase section */
|
||||
|
||||
/* Begin PBXFileReference section */
|
||||
9237521E240C550B00DD6076 /* VoskApiTest.app */ = {isa = PBXFileReference; explicitFileType = wrapper.application; includeInIndex = 0; path = VoskApiTest.app; sourceTree = BUILT_PRODUCTS_DIR; };
|
||||
92375221240C550B00DD6076 /* AppDelegate.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = AppDelegate.swift; sourceTree = "<group>"; };
|
||||
@@ -78,22 +33,11 @@
|
||||
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>"; };
|
||||
92D86BD3253F823E0040D53F /* vosk-model-small-en-us-0.4 */ = {isa = PBXFileReference; lastKnownFileType = folder; path = "vosk-model-small-en-us-0.4"; sourceTree = "<group>"; };
|
||||
92D86BD4253F823F0040D53F /* vosk-model-spk-0.4 */ = {isa = PBXFileReference; lastKnownFileType = folder; path = "vosk-model-spk-0.4"; sourceTree = "<group>"; };
|
||||
/* End PBXFileReference section */
|
||||
|
||||
/* Begin PBXFrameworksBuildPhase section */
|
||||
@@ -146,7 +90,8 @@
|
||||
92375239240C642000DD6076 /* Vosk */ = {
|
||||
isa = PBXGroup;
|
||||
children = (
|
||||
92375247240C6E3D00DD6076 /* model-android */,
|
||||
92D86BD3253F823E0040D53F /* vosk-model-small-en-us-0.4 */,
|
||||
92D86BD4253F823F0040D53F /* vosk-model-spk-0.4 */,
|
||||
92375256240C6E3D00DD6076 /* 10001-90210-01803.wav */,
|
||||
92AA22AD244CDD1200DA464B /* vosk_api.h */,
|
||||
9237523A240C642000DD6076 /* libkaldiwrap.a */,
|
||||
@@ -164,34 +109,6 @@
|
||||
name = Frameworks;
|
||||
sourceTree = "<group>";
|
||||
};
|
||||
92375247240C6E3D00DD6076 /* model-android */ = {
|
||||
isa = PBXGroup;
|
||||
children = (
|
||||
92375248240C6E3D00DD6076 /* disambig_tid.int */,
|
||||
92375249240C6E3D00DD6076 /* final.mdl */,
|
||||
9237524A240C6E3D00DD6076 /* Gr.fst */,
|
||||
9237524B240C6E3D00DD6076 /* HCLr.fst */,
|
||||
9237524C240C6E3D00DD6076 /* ivector */,
|
||||
92375253240C6E3D00DD6076 /* mfcc.conf */,
|
||||
92375254240C6E3D00DD6076 /* word_boundary.int */,
|
||||
92375255240C6E3D00DD6076 /* words.txt */,
|
||||
);
|
||||
path = "model-android";
|
||||
sourceTree = "<group>";
|
||||
};
|
||||
9237524C240C6E3D00DD6076 /* ivector */ = {
|
||||
isa = PBXGroup;
|
||||
children = (
|
||||
9237524D240C6E3D00DD6076 /* final.dubm */,
|
||||
9237524E240C6E3D00DD6076 /* final.ie */,
|
||||
9237524F240C6E3D00DD6076 /* final.mat */,
|
||||
92375250240C6E3D00DD6076 /* global_cmvn.stats */,
|
||||
92375251240C6E3D00DD6076 /* online_cmvn.conf */,
|
||||
92375252240C6E3D00DD6076 /* splice.conf */,
|
||||
);
|
||||
path = ivector;
|
||||
sourceTree = "<group>";
|
||||
};
|
||||
/* End PBXGroup section */
|
||||
|
||||
/* Begin PBXNativeTarget section */
|
||||
@@ -202,8 +119,6 @@
|
||||
9237521A240C550B00DD6076 /* Sources */,
|
||||
9237521B240C550B00DD6076 /* Frameworks */,
|
||||
9237521C240C550B00DD6076 /* Resources */,
|
||||
92375265240C6ECF00DD6076 /* CopyFiles */,
|
||||
9237526D240C6F0400DD6076 /* CopyFiles */,
|
||||
);
|
||||
buildRules = (
|
||||
);
|
||||
@@ -257,7 +172,9 @@
|
||||
92375274240C6F1E00DD6076 /* 10001-90210-01803.wav in Resources */,
|
||||
9237522C240C550B00DD6076 /* LaunchScreen.storyboard in Resources */,
|
||||
92375229240C550B00DD6076 /* Assets.xcassets in Resources */,
|
||||
92D86BD6253F823F0040D53F /* vosk-model-spk-0.4 in Resources */,
|
||||
92375227240C550B00DD6076 /* Main.storyboard in Resources */,
|
||||
92D86BD5253F823F0040D53F /* vosk-model-small-en-us-0.4 in Resources */,
|
||||
);
|
||||
runOnlyForDeploymentPostprocessing = 0;
|
||||
};
|
||||
@@ -412,7 +329,7 @@
|
||||
buildSettings = {
|
||||
ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
|
||||
CLANG_ENABLE_MODULES = YES;
|
||||
ENABLE_BITCODE = NO;
|
||||
ENABLE_BITCODE = YES;
|
||||
INFOPLIST_FILE = VoskApiTest/Info.plist;
|
||||
LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
|
||||
LIBRARY_SEARCH_PATHS = (
|
||||
@@ -434,7 +351,7 @@
|
||||
buildSettings = {
|
||||
ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
|
||||
CLANG_ENABLE_MODULES = YES;
|
||||
ENABLE_BITCODE = NO;
|
||||
ENABLE_BITCODE = YES;
|
||||
INFOPLIST_FILE = VoskApiTest/Info.plist;
|
||||
LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
|
||||
LIBRARY_SEARCH_PATHS = (
|
||||
|
||||
@@ -2,6 +2,12 @@
|
||||
<Workspace
|
||||
version = "1.0">
|
||||
<FileRef
|
||||
location = "self:VoskApiTest.xcodeproj">
|
||||
location = "group:../VoskApiTest/Vosk/vosk-model-small-en-us-0.4">
|
||||
</FileRef>
|
||||
<FileRef
|
||||
location = "group:../VoskApiTest/Vosk/vosk-model-spk-0.4">
|
||||
</FileRef>
|
||||
<FileRef
|
||||
location = "self:">
|
||||
</FileRef>
|
||||
</Workspace>
|
||||
|
||||
@@ -14,10 +14,15 @@ public final class Vosk {
|
||||
var sres = ""
|
||||
if let resourcePath = Bundle.main.resourcePath {
|
||||
|
||||
let modelPath = resourcePath + "/model-en"
|
||||
// Set to -1 to disable logs
|
||||
vosk_set_log_level(0);
|
||||
|
||||
let model = vosk_model_new(modelPath);
|
||||
let recognizer = vosk_recognizer_new(model, 16000.0)
|
||||
let modelPath = resourcePath + "/vosk-model-small-en-us-0.4"
|
||||
let spkModelPath = resourcePath + "/vosk-model-spk-0.4"
|
||||
|
||||
let model = vosk_model_new(modelPath)
|
||||
let spkModel = vosk_spk_model_new(spkModelPath)
|
||||
let recognizer = vosk_recognizer_new_spk(model, spkModel, 16000.0)
|
||||
|
||||
let audioFile = URL(fileURLWithPath: resourcePath + "/10001-90210-01803.wav")
|
||||
if let data = try? Data(contentsOf: audioFile) {
|
||||
|
||||
@@ -88,10 +88,10 @@ VoskRecognizer *vosk_recognizer_new(VoskModel *model, float sample_rate);
|
||||
VoskRecognizer *vosk_recognizer_new_spk(VoskModel *model, VoskSpkModel *spk_model, float sample_rate);
|
||||
|
||||
|
||||
/** Creates the recognizer object with the grammar
|
||||
/** Creates the recognizer object with the phrase list
|
||||
*
|
||||
* Sometimes when you want to improve recognition accuracy and when you don't need
|
||||
* to recognize large vocabulary you can specify a list of words to recognize. This
|
||||
* to recognize large vocabulary you can specify a list of phrases to recognize. This
|
||||
* will improve recognizer speed and accuracy but might return [unk] if user said
|
||||
* something different.
|
||||
*
|
||||
@@ -99,7 +99,8 @@ VoskRecognizer *vosk_recognizer_new_spk(VoskModel *model, VoskSpkModel *spk_mode
|
||||
* Precompiled HCLG graph models are not supported.
|
||||
*
|
||||
* @param sample_rate The sample rate of the audio you going to feed into the recognizer
|
||||
* @param grammar The string with the list of words to recognize, for example "one two three four five [unk]"
|
||||
* @param grammar The string with the list of phrases to recognize as JSON array of strings,
|
||||
* for example "["one two three four five", "[unk]"]".
|
||||
*
|
||||
* @returns recognizer object */
|
||||
VoskRecognizer *vosk_recognizer_new_grm(VoskModel *model, float sample_rate, const char *grammar);
|
||||
|
||||
+15
-10
@@ -1,5 +1,5 @@
|
||||
KALDI_ROOT ?= $(HOME)/kaldi
|
||||
CFLAGS := -g -O2 -DPIC -fPIC -Wno-unused-function
|
||||
CFLAGS := -g -O2 -DPIC -fPIC -Wno-unused-function -std=c++17
|
||||
CPPFLAGS := -I$(JAVA_HOME)/include -I$(JAVA_HOME)/include/linux -I$(KALDI_ROOT)/src -I$(KALDI_ROOT)/tools/openfst/include -I../src
|
||||
|
||||
KALDI_LIBS = \
|
||||
@@ -21,8 +21,13 @@ KALDI_LIBS = \
|
||||
${KALDI_ROOT}/src/base/kaldi-base.a \
|
||||
${KALDI_ROOT}/tools/openfst/lib/libfst.a \
|
||||
${KALDI_ROOT}/tools/openfst/lib/libfstngram.a \
|
||||
${KALDI_ROOT}/tools/OpenBLAS/libopenblas.a \
|
||||
-lgfortran
|
||||
|
||||
# On Linux
|
||||
MATH_LIBS = \
|
||||
${KALDI_ROOT}/tools/OpenBLAS/install/lib/libopenblas.a \
|
||||
${KALDI_ROOT}/tools/OpenBLAS/install/lib/liblapack.a \
|
||||
${KALDI_ROOT}/tools/OpenBLAS/install/lib/libblas.a \
|
||||
${KALDI_ROOT}/tools/OpenBLAS/install/lib/libf2c.a
|
||||
|
||||
all: libvosk_jni.so
|
||||
|
||||
@@ -40,7 +45,7 @@ VOSK_SOURCES = \
|
||||
../src/vosk_api.h
|
||||
|
||||
libvosk_jni.so: $(VOSK_SOURCES)
|
||||
$(CXX) -shared -o $@ $(CPPFLAGS) $(CFLAGS) $(VOSK_SOURCES) $(KALDI_LIBS)
|
||||
$(CXX) -shared -o $@ $(CPPFLAGS) $(CFLAGS) $(VOSK_SOURCES) $(KALDI_LIBS) $(MATH_LIBS)
|
||||
|
||||
vosk_wrap.cc: ../src/vosk.i
|
||||
mkdir -p org/kaldi
|
||||
@@ -53,14 +58,14 @@ clean:
|
||||
$(RM) -r org model-en
|
||||
|
||||
model:
|
||||
wget https://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.zip
|
||||
unzip vosk-model-small-en-us-0.3.zip && rm vosk-model-small-en-us-0.3.zip
|
||||
mv vosk-model-small-en-us-0.3 model
|
||||
wget https://alphacephei.com/kaldi/models/vosk-model-small-en-us-0.3.15.zip
|
||||
unzip vosk-model-small-en-us-0.3.zip && rm vosk-model-small-en-us-0.15.zip
|
||||
mv vosk-model-small-en-us-0.3.15 model
|
||||
|
||||
model-spk:
|
||||
wget https://alphacephei.com/kaldi/models/vosk-model-spk-0.3.zip
|
||||
unzip vosk-model-spk-0.3.zip && rm vosk-model-spk-0.3.zip
|
||||
mv vosk-model-spk-0.3 model-spk
|
||||
wget https://alphacephei.com/kaldi/models/vosk-model-spk-0.4.zip
|
||||
unzip vosk-model-spk-0.3.zip && rm vosk-model-spk-0.4.zip
|
||||
mv vosk-model-spk-0.4 model-spk
|
||||
|
||||
run: model model-spk
|
||||
javac test/*.java org/kaldi/*.java
|
||||
|
||||
+7
-3
@@ -11,7 +11,7 @@
|
||||
'vosk_wrap.cc',
|
||||
],
|
||||
'cflags': [
|
||||
'-std=c++11',
|
||||
'-std=c++17',
|
||||
'-DFST_NO_DYNAMIC_LINKING',
|
||||
'-Wno-deprecated-declarations',
|
||||
'-Wno-sign-compare',
|
||||
@@ -22,13 +22,14 @@
|
||||
'cflags_cc!' : [
|
||||
'-fno-rtti',
|
||||
'-fno-exceptions',
|
||||
'-std=gnu++1y',
|
||||
],
|
||||
'conditions': [
|
||||
['OS == "mac"', {
|
||||
'xcode_settings': {
|
||||
'GCC_ENABLE_CPP_EXCEPTIONS': 'YES',
|
||||
'GCC_ENABLE_CPP_RTTI': 'YES',
|
||||
'CLANG_CXX_LANGUAGE_STANDARD': 'c++11'
|
||||
'CLANG_CXX_LANGUAGE_STANDARD': 'c++17'
|
||||
}
|
||||
}]
|
||||
],
|
||||
@@ -69,7 +70,10 @@
|
||||
'<@(kaldi_root)/src/base/kaldi-base.a',
|
||||
'<@(kaldi_root)/tools/openfst/lib/libfst.a',
|
||||
'<@(kaldi_root)/tools/openfst/lib/libfstngram.a',
|
||||
'<@(kaldi_root)/tools/OpenBLAS/libopenblas.a',
|
||||
'<@(kaldi_root)/tools/OpenBLAS/install/lib/libopenblas.a',
|
||||
'<@(kaldi_root)/tools/OpenBLAS/install/lib/liblapack.a',
|
||||
'<@(kaldi_root)/tools/OpenBLAS/install/lib/libblas.a',
|
||||
'<@(kaldi_root)/tools/OpenBLAS/install/lib/libf2c.a',
|
||||
],
|
||||
'library_dirs': [
|
||||
'/usr/lib',
|
||||
|
||||
+1
-1
@@ -1,6 +1,6 @@
|
||||
{
|
||||
"name": "vosk",
|
||||
"version": "0.3.15",
|
||||
"version": "0.3.17",
|
||||
"description": "Node binding for continuous offline voice recoginition with Vosk library.",
|
||||
"repository": {
|
||||
"type": "git",
|
||||
|
||||
+2
-2
@@ -1,9 +1,9 @@
|
||||
This is a Python module for Vosk.
|
||||
|
||||
Vosk is an offline open source speech recognition toolkit. It enables
|
||||
speech recognition models for 16 languages and dialects - English, Indian
|
||||
speech recognition models for 17 languages and dialects - English, Indian
|
||||
English, German, French, Spanish, Portuguese, Chinese, Russian, Turkish,
|
||||
Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi.
|
||||
Vietnamese, Italian, Dutch, Catalan, Arabic, Greek, Farsi, Filipino.
|
||||
|
||||
Vosk models are small (50 Mb) but provide continuous large vocabulary
|
||||
transcription, zero-latency response with streaming API, reconfigurable
|
||||
|
||||
@@ -44,13 +44,15 @@ while True:
|
||||
if rec.AcceptWaveform(data):
|
||||
res = json.loads(rec.Result())
|
||||
print ("Text:", res['text'])
|
||||
print ("X-vector:", res['spk'])
|
||||
print ("Speaker distance:", cosine_dist(spk_sig, res['spk']), "based on", res['spk_frames'], "frames")
|
||||
if 'spk' in res:
|
||||
print ("X-vector:", res['spk'])
|
||||
print ("Speaker distance:", cosine_dist(spk_sig, res['spk']), "based on", res['spk_frames'], "frames")
|
||||
|
||||
print ("Note that second distance is not very reliable because utterance is too short. Utterances longer than 4 seconds give better xvector")
|
||||
|
||||
res = json.loads(rec.FinalResult())
|
||||
print ("Text:", res['text'])
|
||||
print ("X-vector:", res['spk'])
|
||||
print ("Speaker distance:", cosine_dist(spk_sig, res['spk']), "based on", res['spk_frames'], "frames")
|
||||
if 'spk' in res:
|
||||
print ("X-vector:", res['spk'])
|
||||
print ("Speaker distance:", cosine_dist(spk_sig, res['spk']), "based on", res['spk_frames'], "frames")
|
||||
|
||||
|
||||
@@ -25,6 +25,8 @@ process = subprocess.Popen(['ffmpeg', '-loglevel', 'quiet', '-i',
|
||||
stdout=subprocess.PIPE)
|
||||
|
||||
|
||||
WORDS_PER_LINE = 7
|
||||
|
||||
def transcribe():
|
||||
results = []
|
||||
subs = []
|
||||
@@ -38,12 +40,16 @@ def transcribe():
|
||||
|
||||
for i, res in enumerate(results):
|
||||
jres = json.loads(res)
|
||||
s = srt.Subtitle(index=i,
|
||||
content=jres['text'],
|
||||
start=datetime.timedelta(seconds=jres['result'][0]['start']),
|
||||
end=datetime.timedelta(seconds=jres['result'][-1]['end']))
|
||||
subs.append(s)
|
||||
if not 'result' in jres:
|
||||
continue
|
||||
words = jres['result']
|
||||
for j in range(0, len(words), WORDS_PER_LINE):
|
||||
line = words[j : j + WORDS_PER_LINE]
|
||||
s = srt.Subtitle(index=len(subs),
|
||||
content=" ".join([l['word'] for l in line]),
|
||||
start=datetime.timedelta(seconds=line[0]['start']),
|
||||
end=datetime.timedelta(seconds=line[-1]['end']))
|
||||
subs.append(s)
|
||||
return subs
|
||||
|
||||
print (srt.compose(transcribe()))
|
||||
|
||||
|
||||
+7
-5
@@ -47,7 +47,7 @@ kaldi_static_libs = ['src/online2/kaldi-online2.a',
|
||||
'src/base/kaldi-base.a',
|
||||
'tools/openfst/lib/libfst.a',
|
||||
'tools/openfst/lib/libfstngram.a']
|
||||
kaldi_link_args = ['-s']
|
||||
kaldi_link_args = ['-s', '-latomic', '-lpthread', '-lm']
|
||||
kaldi_libraries = []
|
||||
|
||||
if sys.platform.startswith('darwin'):
|
||||
@@ -56,8 +56,10 @@ elif kaldi_mkl == "1":
|
||||
kaldi_link_args.extend(['-L/opt/intel/mkl/lib/intel64', '-Wl,-rpath=/opt/intel/mkl/lib/intel64'])
|
||||
kaldi_libraries.extend(['mkl_rt', 'mkl_intel_lp64', 'mkl_core', 'mkl_sequential'])
|
||||
else:
|
||||
kaldi_static_libs.append('tools/OpenBLAS/libopenblas.a')
|
||||
kaldi_libraries.append('gfortran')
|
||||
kaldi_static_libs.extend(['tools/OpenBLAS/install/lib/libopenblas.a',
|
||||
'tools/OpenBLAS/install/lib/liblapack.a',
|
||||
'tools/OpenBLAS/install/lib/libblas.a',
|
||||
'tools/OpenBLAS/install/lib/libf2c.a'])
|
||||
|
||||
sources = ['kaldi_recognizer.cc', 'model.cc', 'spk_model.cc', 'vosk_api.cc', 'language_model.cc', 'vosk.i']
|
||||
|
||||
@@ -69,11 +71,11 @@ vosk_ext = Extension('vosk._vosk',
|
||||
extra_objects = [kaldi_root + '/' + x for x in kaldi_static_libs],
|
||||
sources = ['vosk/' + x for x in sources],
|
||||
extra_link_args = kaldi_link_args,
|
||||
extra_compile_args = ['-std=c++11', '-Wno-sign-compare', '-Wno-unused-variable', '-Wno-unused-local-typedefs'])
|
||||
extra_compile_args = ['-O3', '-std=c++17', '-Wno-sign-compare', '-Wno-unused-variable', '-Wno-unused-local-typedefs'])
|
||||
|
||||
setuptools.setup(
|
||||
name="vosk",
|
||||
version="0.3.15",
|
||||
version="0.3.17",
|
||||
author="Alpha Cephei Inc",
|
||||
author_email="contact@alphacephei.com",
|
||||
description="Offline open source speech recognition API based on Kaldi and Vosk",
|
||||
|
||||
+27
-27
@@ -44,7 +44,7 @@ namespace {
|
||||
case '\t': output += "\\t"; break;
|
||||
default : output += str[i]; break;
|
||||
}
|
||||
return std::move( output );
|
||||
return output;
|
||||
}
|
||||
}
|
||||
|
||||
@@ -291,10 +291,10 @@ class JSON
|
||||
/// Functions for getting primitives from the JSON object.
|
||||
bool IsNull() const { return Type == Class::Null; }
|
||||
|
||||
string ToString() const { bool b; return std::move( ToString( b ) ); }
|
||||
string ToString() const { bool b; return ToString(b); }
|
||||
string ToString( bool &ok ) const {
|
||||
ok = (Type == Class::String);
|
||||
return ok ? std::move( json_escape( *Internal.String ) ): string("");
|
||||
return ok ? json_escape( *Internal.String ) : string("");
|
||||
}
|
||||
|
||||
double ToFloat() const { bool b; return ToFloat( b ); }
|
||||
@@ -425,18 +425,18 @@ class JSON
|
||||
};
|
||||
|
||||
JSON Array() {
|
||||
return std::move( JSON::Make( JSON::Class::Array ) );
|
||||
return JSON::Make( JSON::Class::Array );
|
||||
}
|
||||
|
||||
template <typename... T>
|
||||
JSON Array( T... args ) {
|
||||
JSON arr = JSON::Make( JSON::Class::Array );
|
||||
arr.append( args... );
|
||||
return std::move( arr );
|
||||
return arr;
|
||||
}
|
||||
|
||||
JSON Object() {
|
||||
return std::move( JSON::Make( JSON::Class::Object ) );
|
||||
return JSON::Make( JSON::Class::Object );
|
||||
}
|
||||
|
||||
std::ostream& operator<<( std::ostream &os, const JSON &json ) {
|
||||
@@ -457,7 +457,7 @@ namespace {
|
||||
++offset;
|
||||
consume_ws( str, offset );
|
||||
if( str[offset] == '}' ) {
|
||||
++offset; return std::move( Object );
|
||||
++offset; return Object;
|
||||
}
|
||||
|
||||
while( true ) {
|
||||
@@ -484,7 +484,7 @@ namespace {
|
||||
}
|
||||
}
|
||||
|
||||
return std::move( Object );
|
||||
return Object;
|
||||
}
|
||||
|
||||
JSON parse_array( const string &str, size_t &offset ) {
|
||||
@@ -494,7 +494,7 @@ namespace {
|
||||
++offset;
|
||||
consume_ws( str, offset );
|
||||
if( str[offset] == ']' ) {
|
||||
++offset; return std::move( Array );
|
||||
++offset; return Array;
|
||||
}
|
||||
|
||||
while( true ) {
|
||||
@@ -509,11 +509,11 @@ namespace {
|
||||
}
|
||||
else {
|
||||
std::cerr << "ERROR: Array: Expected ',' or ']', found '" << str[offset] << "'\n";
|
||||
return std::move( JSON::Make( JSON::Class::Array ) );
|
||||
return JSON::Make( JSON::Class::Array );
|
||||
}
|
||||
}
|
||||
|
||||
return std::move( Array );
|
||||
return Array;
|
||||
}
|
||||
|
||||
JSON parse_string( const string &str, size_t &offset ) {
|
||||
@@ -538,7 +538,7 @@ namespace {
|
||||
val += c;
|
||||
else {
|
||||
std::cerr << "ERROR: String: Expected hex character in unicode escape, found '" << c << "'\n";
|
||||
return std::move( JSON::Make( JSON::Class::String ) );
|
||||
return JSON::Make( JSON::Class::String );
|
||||
}
|
||||
}
|
||||
offset += 4;
|
||||
@@ -551,7 +551,7 @@ namespace {
|
||||
}
|
||||
++offset;
|
||||
String = val;
|
||||
return std::move( String );
|
||||
return String;
|
||||
}
|
||||
|
||||
JSON parse_number( const string &str, size_t &offset ) {
|
||||
@@ -580,7 +580,7 @@ namespace {
|
||||
exp_str += c;
|
||||
else if( !isspace( c ) && c != ',' && c != ']' && c != '}' ) {
|
||||
std::cerr << "ERROR: Number: Expected a number for exponent, found '" << c << "'\n";
|
||||
return std::move( JSON::Make( JSON::Class::Null ) );
|
||||
return JSON::Make( JSON::Class::Null );
|
||||
}
|
||||
else
|
||||
break;
|
||||
@@ -589,7 +589,7 @@ namespace {
|
||||
}
|
||||
else if( !isspace( c ) && c != ',' && c != ']' && c != '}' ) {
|
||||
std::cerr << "ERROR: Number: unexpected character '" << c << "'\n";
|
||||
return std::move( JSON::Make( JSON::Class::Null ) );
|
||||
return JSON::Make( JSON::Class::Null );
|
||||
}
|
||||
--offset;
|
||||
|
||||
@@ -601,7 +601,7 @@ namespace {
|
||||
else
|
||||
Number = std::stol( val );
|
||||
}
|
||||
return std::move( Number );
|
||||
return Number;
|
||||
}
|
||||
|
||||
JSON parse_bool( const string &str, size_t &offset ) {
|
||||
@@ -612,20 +612,20 @@ namespace {
|
||||
Bool = false;
|
||||
else {
|
||||
std::cerr << "ERROR: Bool: Expected 'true' or 'false', found '" << str.substr( offset, 5 ) << "'\n";
|
||||
return std::move( JSON::Make( JSON::Class::Null ) );
|
||||
return JSON::Make( JSON::Class::Null );
|
||||
}
|
||||
offset += (Bool.ToBool() ? 4 : 5);
|
||||
return std::move( Bool );
|
||||
return Bool;
|
||||
}
|
||||
|
||||
JSON parse_null( const string &str, size_t &offset ) {
|
||||
JSON Null;
|
||||
if( str.substr( offset, 4 ) != "null" ) {
|
||||
std::cerr << "ERROR: Null: Expected 'null', found '" << str.substr( offset, 4 ) << "'\n";
|
||||
return std::move( JSON::Make( JSON::Class::Null ) );
|
||||
return JSON::Make( JSON::Class::Null );
|
||||
}
|
||||
offset += 4;
|
||||
return std::move( Null );
|
||||
return Null;
|
||||
}
|
||||
|
||||
JSON parse_next( const string &str, size_t &offset ) {
|
||||
@@ -633,14 +633,14 @@ namespace {
|
||||
consume_ws( str, offset );
|
||||
value = str[offset];
|
||||
switch( value ) {
|
||||
case '[' : return std::move( parse_array( str, offset ) );
|
||||
case '{' : return std::move( parse_object( str, offset ) );
|
||||
case '\"': return std::move( parse_string( str, offset ) );
|
||||
case '[' : return parse_array( str, offset );
|
||||
case '{' : return parse_object( str, offset );
|
||||
case '\"': return parse_string( str, offset );
|
||||
case 't' :
|
||||
case 'f' : return std::move( parse_bool( str, offset ) );
|
||||
case 'n' : return std::move( parse_null( str, offset ) );
|
||||
case 'f' : return parse_bool( str, offset );
|
||||
case 'n' : return parse_null( str, offset );
|
||||
default : if( ( value <= '9' && value >= '0' ) || value == '-' )
|
||||
return std::move( parse_number( str, offset ) );
|
||||
return parse_number( str, offset );
|
||||
}
|
||||
std::cerr << "ERROR: Parse: Unknown starting character '" << value << "'\n";
|
||||
return JSON();
|
||||
@@ -649,7 +649,7 @@ namespace {
|
||||
|
||||
JSON JSON::Load( const string &str ) {
|
||||
size_t offset = 0;
|
||||
return std::move( parse_next( str, offset ) );
|
||||
return parse_next( str, offset );
|
||||
}
|
||||
|
||||
} // End Namespace json
|
||||
|
||||
@@ -173,9 +173,9 @@ void KaldiRecognizer::InitRescoring()
|
||||
{
|
||||
if (model_->std_lm_fst_) {
|
||||
fst::CacheOptions cache_opts(true, 50000);
|
||||
fst::MapFstOptions mapfst_opts(cache_opts);
|
||||
fst::ArcMapFstOptions 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);
|
||||
lm_fst_ = new fst::ArcMapFst<fst::StdArc, kaldi::LatticeArc, fst::StdToLatticeMapper<kaldi::BaseFloat> >(*model_->std_lm_fst_, mapper, mapfst_opts);
|
||||
} else {
|
||||
lm_fst_ = NULL;
|
||||
}
|
||||
|
||||
@@ -72,7 +72,7 @@ class KaldiRecognizer {
|
||||
SpkModel *spk_model_;
|
||||
OnlineBaseFeature *spk_feature_;
|
||||
|
||||
fst::MapFst<fst::StdArc, kaldi::LatticeArc, fst::StdToLatticeMapper<kaldi::BaseFloat> > *lm_fst_;
|
||||
fst::ArcMapFst<fst::StdArc, kaldi::LatticeArc, fst::StdToLatticeMapper<kaldi::BaseFloat> > *lm_fst_;
|
||||
|
||||
float sample_frequency_;
|
||||
int32 frame_offset_;
|
||||
|
||||
@@ -161,6 +161,7 @@ void Model::ConfigureV1()
|
||||
std_fst_rxfilename_ = model_path_str_ + "/rescore/G.fst";
|
||||
final_ie_rxfilename_ = model_path_str_ + "/ivector/final.ie";
|
||||
mfcc_conf_rxfilename_ = model_path_str_ + "/mfcc.conf";
|
||||
global_cmvn_stats_rxfilename_ = model_path_str_ + "/global_cmvn.stats";
|
||||
}
|
||||
|
||||
void Model::ConfigureV2()
|
||||
@@ -183,6 +184,7 @@ void Model::ConfigureV2()
|
||||
std_fst_rxfilename_ = model_path_str_ + "/rescore/G.fst";
|
||||
final_ie_rxfilename_ = model_path_str_ + "/ivector/final.ie";
|
||||
mfcc_conf_rxfilename_ = model_path_str_ + "/conf/mfcc.conf";
|
||||
global_cmvn_stats_rxfilename_ = model_path_str_ + "/am/global_cmvn.stats";
|
||||
}
|
||||
|
||||
void Model::ReadDataFiles()
|
||||
@@ -233,6 +235,13 @@ void Model::ReadDataFiles()
|
||||
feature_info_.use_ivectors = false;
|
||||
}
|
||||
|
||||
if (stat(global_cmvn_stats_rxfilename_.c_str(), &buffer) == 0) {
|
||||
KALDI_LOG << "Reading CMVN stats from " << global_cmvn_stats_rxfilename_;
|
||||
feature_info_.use_cmvn = true;
|
||||
ReadKaldiObject(global_cmvn_stats_rxfilename_, &feature_info_.global_cmvn_stats);
|
||||
}
|
||||
|
||||
|
||||
if (stat(hclg_fst_rxfilename_.c_str(), &buffer) == 0) {
|
||||
KALDI_LOG << "Loading HCLG from " << hclg_fst_rxfilename_;
|
||||
hclg_fst_ = fst::ReadFstKaldiGeneric(hclg_fst_rxfilename_);
|
||||
@@ -295,6 +304,14 @@ void Model::Unref()
|
||||
}
|
||||
}
|
||||
|
||||
int Model::FindWord(const char *word)
|
||||
{
|
||||
if (!word_syms_)
|
||||
return -1;
|
||||
|
||||
return word_syms_->Find(word);
|
||||
}
|
||||
|
||||
Model::~Model() {
|
||||
delete decodable_info_;
|
||||
delete trans_model_;
|
||||
|
||||
@@ -42,6 +42,7 @@ public:
|
||||
Model(const char *model_path);
|
||||
void Ref();
|
||||
void Unref();
|
||||
int FindWord(const char *word);
|
||||
|
||||
protected:
|
||||
~Model();
|
||||
@@ -63,6 +64,7 @@ protected:
|
||||
string std_fst_rxfilename_;
|
||||
string final_ie_rxfilename_;
|
||||
string mfcc_conf_rxfilename_;
|
||||
string global_cmvn_stats_rxfilename_;
|
||||
|
||||
kaldi::OnlineEndpointConfig endpoint_config_;
|
||||
kaldi::LatticeFasterDecoderConfig nnet3_decoding_config_;
|
||||
|
||||
@@ -75,6 +75,9 @@ typedef struct {} KaldiRecognizer;
|
||||
~Model() {
|
||||
vosk_model_free($self);
|
||||
}
|
||||
int vosk_model_find_word(const char* word) {
|
||||
return vosk_model_find_word($self, word);
|
||||
}
|
||||
}
|
||||
|
||||
%extend SpkModel {
|
||||
|
||||
@@ -31,6 +31,11 @@ void vosk_model_free(VoskModel *model)
|
||||
((Model *)model)->Unref();
|
||||
}
|
||||
|
||||
int vosk_model_find_word(VoskModel *model, const char *word)
|
||||
{
|
||||
return (int) ((Model *)model)->FindWord(word);
|
||||
}
|
||||
|
||||
VoskSpkModel *vosk_spk_model_new(const char *model_path)
|
||||
{
|
||||
return (VoskSpkModel *)new SpkModel(model_path);
|
||||
|
||||
@@ -55,6 +55,14 @@ VoskModel *vosk_model_new(const char *model_path);
|
||||
void vosk_model_free(VoskModel *model);
|
||||
|
||||
|
||||
/** Check if a word can be recognized by the model
|
||||
* @param word: the word
|
||||
* @returns the word symbol if @param word exists inside the model
|
||||
* or -1 otherwise.
|
||||
* Reminding that word symbol 0 is for <epsilon> */
|
||||
int vosk_model_find_word(VoskModel *model, const char *word);
|
||||
|
||||
|
||||
/** Loads speaker model data from the file and returns the model object
|
||||
*
|
||||
* @param model_path: the path of the model on the filesystem
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
ARG DOCKCROSS_IMAGE=linux-armv7
|
||||
FROM dockcross/${DOCKCROSS_IMAGE}
|
||||
ARG DOCKCROSS_IMAGE=alphacep/dockcross-linux-armv7
|
||||
FROM ${DOCKCROSS_IMAGE}
|
||||
|
||||
LABEL description="A docker image for building portable Python linux binary wheels and Kaldi on other architectures"
|
||||
LABEL maintainer="contact@alphacephei.com"
|
||||
@@ -13,6 +13,7 @@ RUN apt-get update && \
|
||||
automake \
|
||||
autoconf \
|
||||
libtool \
|
||||
cmake \
|
||||
&& rm -rf /var/lib/apt/lists/*
|
||||
|
||||
RUN cd /opt \
|
||||
@@ -86,19 +87,23 @@ RUN cd /opt \
|
||||
ARG OPENBLAS_ARCH=ARMV7
|
||||
ARG ARM_HARDWARE_OPTS="-mfloat-abi=hard -mfpu=neon"
|
||||
RUN cd /opt \
|
||||
&& git clone -b lookahead --single-branch https://github.com/alphacep/kaldi \
|
||||
&& git clone -b lookahead-1.8.0 --single-branch https://github.com/alphacep/kaldi \
|
||||
&& cd kaldi/tools \
|
||||
&& git clone -b v0.3.7 --single-branch https://github.com/xianyi/OpenBLAS \
|
||||
&& make PREFIX=$(pwd)/OpenBLAS/install TARGET="${OPENBLAS_ARCH}" HOSTCC=gcc USE_LOCKING=1 USE_THREAD=0 -C OpenBLAS all install \
|
||||
&& git clone -b old-gcc --single-branch https://github.com/alphacep/openfst openfst \
|
||||
&& git clone -b v0.3.13 --single-branch https://github.com/xianyi/OpenBLAS \
|
||||
&& git clone -b v3.2.1 --single-branch https://github.com/alphacep/clapack \
|
||||
&& make -C OpenBLAS ONLY_CBLAS=1 TARGET="${OPENBLAS_ARCH}" HOSTCC=gcc USE_LOCKING=1 USE_THREAD=0 all \
|
||||
&& make -C OpenBLAS PREFIX=$(pwd)/OpenBLAS/install install \
|
||||
&& mkdir -p clapack/BUILD && cd clapack/BUILD && cmake .. && make -j 10 && find . -name "*.a" | xargs cp -t ../../OpenBLAS/install/lib \
|
||||
&& cd /opt/kaldi/tools \
|
||||
&& git clone --single-branch https://github.com/alphacep/openfst openfst \
|
||||
&& cd openfst \
|
||||
&& autoreconf -i \
|
||||
&& ./configure --prefix=/opt/kaldi/tools/openfst --enable-static --enable-shared --enable-far --enable-ngram-fsts --enable-lookahead-fsts --with-pic --disable-bin --host=${CROSS_TRIPLE} --build=x86-linux-gnu \
|
||||
&& CFLAGS="-g -O3" ./configure --prefix=/opt/kaldi/tools/openfst --enable-static --enable-shared --enable-far --enable-ngram-fsts --enable-lookahead-fsts --with-pic --disable-bin --host=${CROSS_TRIPLE} --build=x86-linux-gnu \
|
||||
&& make -j 10 && make install \
|
||||
&& cd /opt/kaldi/src \
|
||||
&& sed -i "s:TARGET_ARCH=\"\`uname -m\`\":TARGET_ARCH=$(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 \
|
||||
&& ./configure --mathlib=OPENBLAS_CLAPACK --shared --use-cuda=no \
|
||||
&& make -j 10 online2 lm \
|
||||
&& find /opt/kaldi -name "*.o" -exec rm {} \;
|
||||
|
||||
@@ -7,9 +7,11 @@ RUN yum -y update && yum -y install \
|
||||
wget \
|
||||
openssl-devel \
|
||||
pcre-devel \
|
||||
devtoolset-8-libatomic-devel \
|
||||
automake \
|
||||
autoconf \
|
||||
libtool \
|
||||
cmake \
|
||||
&& yum clean all
|
||||
|
||||
RUN cd /opt \
|
||||
@@ -21,16 +23,22 @@ RUN cd /opt \
|
||||
&& rm -rf swig-4.0.1.tar.gz swig-4.0.1
|
||||
|
||||
RUN cd /opt \
|
||||
&& git clone -b lookahead --single-branch https://github.com/alphacep/kaldi \
|
||||
&& cd kaldi/tools \
|
||||
&& git clone -b v0.3.7 --single-branch https://github.com/xianyi/OpenBLAS \
|
||||
&& make PREFIX=$(pwd)/OpenBLAS/install DYNAMIC_ARCH=1 USE_LOCKING=1 USE_THREAD=0 -C OpenBLAS all install \
|
||||
&& git clone https://github.com/alphacep/openfst openfst \
|
||||
&& git clone -b lookahead-1.8.0 --single-branch https://github.com/alphacep/kaldi \
|
||||
&& cd /opt/kaldi/tools \
|
||||
&& git clone -b v0.3.13 --single-branch https://github.com/xianyi/OpenBLAS \
|
||||
&& git clone -b v3.2.1 --single-branch https://github.com/alphacep/clapack \
|
||||
&& make -C OpenBLAS ONLY_CBLAS=1 DYNAMIC_ARCH=1 TARGET=NEHALEM USE_LOCKING=1 USE_THREAD=0 all \
|
||||
&& make -C OpenBLAS PREFIX=$(pwd)/OpenBLAS/install install \
|
||||
&& mkdir -p clapack/BUILD && cd clapack/BUILD && cmake .. && make -j 10 && find . -name "*.a" | xargs cp -t ../../OpenBLAS/install/lib \
|
||||
&& cd /opt/kaldi/tools \
|
||||
&& git clone --single-branch https://github.com/alphacep/openfst openfst \
|
||||
&& cd openfst \
|
||||
&& autoreconf -i \
|
||||
&& ./configure --prefix=/opt/kaldi/tools/openfst --enable-static --enable-shared --enable-far --enable-ngram-fsts --enable-lookahead-fsts --with-pic --disable-bin \
|
||||
&& CFLAGS="-g -O3" ./configure --prefix=/opt/kaldi/tools/openfst --enable-static --enable-shared --enable-far --enable-ngram-fsts --enable-lookahead-fsts --with-pic --disable-bin \
|
||||
&& make -j 10 && make install \
|
||||
&& cd /opt/kaldi/src \
|
||||
&& ./configure --mathlib=OPENBLAS --shared --use-cuda=no \
|
||||
&& make -j 10 online2 lm rnnlm \
|
||||
&& ./configure --mathlib=OPENBLAS_CLAPACK --shared --use-cuda=no \
|
||||
&& sed -i 's:-msse -msse2:-msse -msse2:g' kaldi.mk \
|
||||
&& sed -i 's: -O1 : -O3 :g' kaldi.mk \
|
||||
&& make -j $(nproc) online2 lm rnnlm \
|
||||
&& find /opt/kaldi -name "*.o" -exec rm {} \;
|
||||
|
||||
@@ -3,10 +3,8 @@
|
||||
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" --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 build --build-arg="DOCKCROSS_IMAGE=alphacep/dockcross-linux-armv7" --build-arg="OPENBLAS_ARCH=ARMV7" --file Dockerfile.dockcross --tag alphacep/kaldi-dockcross-armv7:latest .
|
||||
docker build --build-arg="DOCKCROSS_IMAGE=dockcross/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
|
||||
|
||||
@@ -13,6 +13,9 @@ for pyver in 3.6 3.7 3.8; do
|
||||
if [ $pyver == "3.8" ]; then
|
||||
sysconfig_bit=""
|
||||
fi
|
||||
if [ $pyver == "3.9" ]; then
|
||||
sysconfig_bit=""
|
||||
fi
|
||||
|
||||
case $CROSS_TRIPLE in
|
||||
*arm-*)
|
||||
|
||||
@@ -4,7 +4,7 @@ set -e -x
|
||||
export KALDI_ROOT=/opt/kaldi
|
||||
|
||||
# Compile wheels
|
||||
for pypath in /opt/python/cp3[56789]*; do
|
||||
for pypath in /opt/python/cp3[6789]*; do
|
||||
export VOSK_SOURCE=/io/src
|
||||
mkdir -p /opt/wheelhouse
|
||||
rm -rf /io/python/build
|
||||
|
||||
Reference in New Issue
Block a user