Bump Emscripten to 3.1.58, bump Vosk to 0.3.50, small refactor and doc.

This commit is contained in:
msqr1
2024-04-23 20:59:04 -07:00
parent 5f5a7a7571
commit 64e4c4c117
10 changed files with 26 additions and 32 deletions

2
API.md
View File

@@ -49,7 +49,7 @@ If you can't set them, you may use a hacky workaround at *src/addCOI.js*.
- Requires ```autotools```'s commands in PATH
- Changing any option to non-default values requires recompilation
- To re-make a specific target, erase its directory in the repo root and run ```./make``` again. Doing this will also re-make the final JS
```
```shell
git clone --depth=1 https://github.com/msqr1/Vosklet &&
cd Vosklet/src &&
[Options] ./make

View File

@@ -3,7 +3,7 @@
- Designed with basic/nothrow exception safety
- See the *examples* folder for examples on using the API.
- See *API.md* for the API reference
- See the *devel* folder for the newest build (not guaranteed to work) and the JS build script
- See the *devel* folder for the build script for development
# Compared to vosk-browser
- Support multiple models
@@ -18,7 +18,7 @@
# Basic usage (microphone recognition)
- Result are logged to the console.
- Copied from *examples/fromMic.html*
```js
```html
<!DOCTYPE html>
<html>
<head>

File diff suppressed because one or more lines are too long

View File

@@ -38,10 +38,10 @@ LIBARCHIVE=$(realpath libarchive)
CLAPACK_WASM=$(realpath clapack-wasm)
cd $(realpath src)
MODE=0 && # 0: Ultra debug info, 1: Extremely optimized release, else custom
MODE=1 && # 0: Ultra debug info, 1: Extremely optimized release, else custom
echo "Mode = $MODE" &&
if [ $MODE = 0 ]; then
em++ -O0 link.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sRUNTIME_DEBUG -sSTACK_OVERFLOW_CHECK=2 -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sASSERTIONS=2 -sINITIAL_MEMORY=$MAX_MEMORY -sPTHREAD_POOL_SIZE=$MAX_THREADS -sDISABLE_EXCEPTION_CATCHING=0 -sEXIT_RUNTIME=0 -sINVOKE_RUN=0 -sPOLYFILL=0 -sEXPORTED_FUNCTIONS=_malloc -sEXPORT_NAME=loadVosklet -sMALLOC=emmalloc -sEXPORTED_RUNTIME_METHODS=UTF8ToString,stringToUTF8OnStack -sENVIRONMENT=web,worker -I. -I$LIBARCHIVE/include -I$VOSK/src -L$LIBARCHIVE/lib -larchive -L$KALDI/src -l:online2/kaldi-online2.a -l:decoder/kaldi-decoder.a -l:ivector/kaldi-ivector.a -l:gmm/kaldi-gmm.a -l:tree/kaldi-tree.a -l:feat/kaldi-feat.a -l:cudamatrix/kaldi-cudamatrix.a -l:lat/kaldi-lat.a -l:lm/kaldi-lm.a -l:rnnlm/kaldi-rnnlm.a -l:hmm/kaldi-hmm.a -l:nnet3/kaldi-nnet3.a -l:transform/kaldi-transform.a -l:matrix/kaldi-matrix.a -l:fstext/kaldi-fstext.a -l:util/kaldi-util.a -l:base/kaldi-base.a -L$OPENFST/lib -l:libfst.a -l:libfstngram.a -L$CLAPACK_WASM -l:CBLAS/lib/cblas.a -l:CLAPACK-3.2.1/lapack.a -l:CLAPACK-3.2.1/libcblaswr.a -l:f2c_BLAS-3.8.0/blas.a -l:libf2c/libf2c.a -L$VOSK/src -l:vosk.a -lembind -pthread -flto -msimd128 --emit-tsd Vosklet.d.ts -fsanitize=undefined -fsanitize=address -fsanitize=leak -msimd128 -mreference-types -mnontrapping-fptoint -mextended-const -msign-ext -mmutable-globals -g3 --pre-js pre.js -o ../Vosklet.js
em++ -O0 link.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sRUNTIME_DEBUG -sSTACK_OVERFLOW_CHECK=2 -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sASSERTIONS=2 -sINITIAL_MEMORY=$MAX_MEMORY -sPTHREAD_POOL_SIZE=$MAX_THREADS -sDISABLE_EXCEPTION_CATCHING=0 -sEXIT_RUNTIME=0 -sINVOKE_RUN=0 -sPOLYFILL=0 -sEXPORTED_FUNCTIONS=_malloc -sEXPORT_NAME=loadVosklet -sMALLOC=emmalloc -sEXPORTED_RUNTIME_METHODS=UTF8ToString,stringToUTF8OnStack -sENVIRONMENT=web,worker -I. -I$LIBARCHIVE/include -I$VOSK/src -L$LIBARCHIVE/lib -larchive -L$KALDI/src -l:online2/kaldi-online2.a -l:decoder/kaldi-decoder.a -l:ivector/kaldi-ivector.a -l:gmm/kaldi-gmm.a -l:tree/kaldi-tree.a -l:feat/kaldi-feat.a -l:cudamatrix/kaldi-cudamatrix.a -l:lat/kaldi-lat.a -l:lm/kaldi-lm.a -l:rnnlm/kaldi-rnnlm.a -l:hmm/kaldi-hmm.a -l:nnet3/kaldi-nnet3.a -l:transform/kaldi-transform.a -l:matrix/kaldi-matrix.a -l:fstext/kaldi-fstext.a -l:util/kaldi-util.a -l:base/kaldi-base.a -L$OPENFST/lib -l:libfst.a -l:libfstngram.a -L$CLAPACK_WASM -l:CBLAS/lib/cblas.a -l:CLAPACK-3.2.1/lapack.a -l:CLAPACK-3.2.1/libcblaswr.a -l:f2c_BLAS-3.8.0/blas.a -l:libf2c/libf2c.a -L$VOSK/src -l:vosk.a -lembind -pthread -flto --emit-tsd Vosklet.d.ts -fsanitize=undefined -fsanitize=address -fsanitize=leak -msimd128 -mreference-types -mnontrapping-fptoint -mextended-const -msign-ext -mmutable-globals -g3 --pre-js pre.js -o ../Vosklet.js
elif [ $MODE = 1 ]; then
em++ -O3 link.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS -sWASM_BIGINT -sSINGLE_FILE -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sINITIAL_MEMORY=$MAX_MEMORY -sPTHREAD_POOL_SIZE=$MAX_THREADS -sPOLYFILL=0 -sEXIT_RUNTIME=0 -sINVOKE_RUN=0 -sSUPPORT_LONGJMP=0 -sEXPORTED_FUNCTIONS=_malloc -sEXPORT_NAME=loadVosklet -sMALLOC=emmalloc -sEXPORTED_RUNTIME_METHODS=UTF8ToString,stringToUTF8OnStack -sENVIRONMENT=web,worker -I. -I$LIBARCHIVE/include -I$VOSK/src -L$LIBARCHIVE/lib -larchive -L$KALDI/src -l:online2/kaldi-online2.a -l:decoder/kaldi-decoder.a -l:ivector/kaldi-ivector.a -l:gmm/kaldi-gmm.a -l:tree/kaldi-tree.a -l:feat/kaldi-feat.a -l:cudamatrix/kaldi-cudamatrix.a -l:lat/kaldi-lat.a -l:lm/kaldi-lm.a -l:rnnlm/kaldi-rnnlm.a -l:hmm/kaldi-hmm.a -l:nnet3/kaldi-nnet3.a -l:transform/kaldi-transform.a -l:matrix/kaldi-matrix.a -l:fstext/kaldi-fstext.a -l:util/kaldi-util.a -l:base/kaldi-base.a -L$OPENFST/lib -l:libfst.a -l:libfstngram.a -L$CLAPACK_WASM -l:CBLAS/lib/cblas.a -l:CLAPACK-3.2.1/lapack.a -l:CLAPACK-3.2.1/libcblaswr.a -l:f2c_BLAS-3.8.0/blas.a -l:libf2c/libf2c.a -L$VOSK/src -l:vosk.a -lembind -pthread -flto -msimd128 -mreference-types -mnontrapping-fptoint -mextended-const -msign-ext -mmutable-globals --pre-js pre.js -o ../Vosklet.js
else

View File

@@ -4,12 +4,13 @@ genericModel::genericModel(int index, bool normalMdl, std::string storepath, std
void genericModel::extractAndLoad(int tarStart, int tarSize) {
static fs::path path{};
static int fd{};
func = [this, tarStart, tarSize](){
void* tar{reinterpret_cast<void*>(tarStart)};
func = [this, tar, tarSize](){
archive* src {archive_read_new()};
archive_read_support_format_tar(src);
archive_read_open_memory(src, reinterpret_cast<void*>(tarStart), tarSize);
archive_read_open_memory(src, tar, tarSize);
if(archive_errno(src) != 0) {
free(reinterpret_cast<void*>(tarStart));
free(tar);
fireEv(index, "Unable to open tar in WASM memory");
return;
}
@@ -38,12 +39,12 @@ void genericModel::extractAndLoad(int tarStart, int tarSize) {
return;
}
}
free(reinterpret_cast<void*>(tarStart));
free(tar);
archive_read_free(src);
if(normalMdl) mdl = vosk_model_new(storepath.c_str());
else vosk_spk_model_new(storepath.c_str());
if(normalMdl ? std::get<0>(mdl) == nullptr : std::get<1>(mdl) == nullptr) fireEv(index, "Unable to load model for recognition");
else fireEv(index, "0");
else mdl = vosk_spk_model_new(storepath.c_str());
if(normalMdl ? std::get<VoskModel*>(mdl) != nullptr : std::get<VoskSpkModel*>(mdl) != nullptr) fireEv(index, "0");
else fireEv(index, "Unable to load model for recognition");
fs::remove_all(storepath);
};
std::thread t{[this](){
@@ -55,6 +56,6 @@ void genericModel::extractAndLoad(int tarStart, int tarSize) {
}
genericModel::~genericModel() {
archive_entry_free(entry);
if(normalMdl) vosk_model_free(std::get<0>(mdl));
else vosk_spk_model_free(std::get<1>(mdl));
if(normalMdl) vosk_model_free(std::get<VoskModel*>(mdl));
else vosk_spk_model_free(std::get<VoskSpkModel*>(mdl));
}

View File

@@ -2,6 +2,7 @@
#include "link.h"
#include <filesystem>
#include <functional>
#include <variant>
#include <fcntl.h>

View File

@@ -1,13 +1,7 @@
#include "link.h"
auto dstThrd = pthread_self();
ProxyingQueue glbQ{};
void fireEv(int index, const char* content, const char* type) {
auto proxy{[index, content, type](){
EM_ASM({
objs[$0].dispatchEvent(new CustomEvent($2 === 0 ? "0" : UTF8ToString($2), { "detail" : UTF8ToString($1) }));
}, index, content, type);
}};
if(dstThrd == pthread_self()) proxy();
else glbQ.proxySync(dstThrd, proxy);
MAIN_THREAD_EM_ASM({
objs[$0].dispatchEvent(new CustomEvent($2 === 0 ? "0" : UTF8ToString($2), { "detail" : UTF8ToString($1) }));
}, index, content, type);
}

View File

@@ -2,7 +2,5 @@
#include <thread>
#include <emscripten/em_asm.h>
#include <emscripten/proxying.h>
using namespace emscripten;
void fireEv(int index, const char* content, const char* type = nullptr); // Normal
void fireEv(int index, const char* content, const char* type = nullptr);

View File

@@ -75,7 +75,7 @@ if [ ! -d $KALDI ]; then
fi
if [ ! -d $VOSK ]; then
git clone -b go/v0.3.46 --depth=1 https://github.com/alphacep/vosk-api $VOSK &&
git clone -b v0.3.50 --depth=1 https://github.com/alphacep/vosk-api $VOSK &&
cd $VOSK/src &&
git apply $SRC/vosk.patch &&
VOSK_FILES="recognizer.cc language_model.cc model.cc spk_model.cc vosk_api.cc" &&

View File

@@ -1,12 +1,12 @@
#include "recognizer.h"
recognizer::recognizer(int index, float sampleRate, genericModel* model) : index{index}, rec{vosk_recognizer_new(std::get<0>(model->mdl),sampleRate)} {
recognizer::recognizer(int index, float sampleRate, genericModel* model) : index{index}, rec{vosk_recognizer_new(std::get<VoskModel*>(model->mdl),sampleRate)} {
finishConstruction(model);
}
recognizer::recognizer(int index, float sampleRate, genericModel* model, genericModel* spkModel) : index{index}, rec{vosk_recognizer_new_spk(std::get<0>(model->mdl), sampleRate, std::get<1>(spkModel->mdl))} {
recognizer::recognizer(int index, float sampleRate, genericModel* model, genericModel* spkModel) : index{index}, rec{vosk_recognizer_new_spk(std::get<VoskModel*>(model->mdl), sampleRate, std::get<VoskSpkModel*>(spkModel->mdl))} {
finishConstruction(model, spkModel);
}
recognizer::recognizer(int index, float sampleRate, genericModel* model, const std::string& grm, int dummy) : index{index}, rec{vosk_recognizer_new_grm(std::get<0>(model->mdl), sampleRate, grm.c_str())} {
recognizer::recognizer(int index, float sampleRate, genericModel* model, const std::string& grm, int dummy) : index{index}, rec{vosk_recognizer_new_grm(std::get<VoskModel*>(model->mdl), sampleRate, grm.c_str())} {
finishConstruction(model);
}
recognizer::~recognizer() {
@@ -71,7 +71,7 @@ void recognizer::setGrm(const std::string& grm) {
vosk_recognizer_set_grm(rec, grm.c_str());
}
void recognizer::setSpkModel(genericModel* spkModel) {
vosk_recognizer_set_spk_model(rec, std::get<1>(spkModel->mdl));
vosk_recognizer_set_spk_model(rec, std::get<VoskSpkModel*>(spkModel->mdl));
}
void recognizer::setWords(bool words) {
vosk_recognizer_set_words(rec,words);