patrick/Vosklet
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Finally

Browse Source
This commit is contained in:
msqr1
2024-03-26 22:04:02 -07:00
parent df6b240bde
commit 7b595ec0ee
21 changed files with 328 additions and 6300 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
src/Makefile
View File

@@ -20,7 +20,7 @@ CLAPACK_WASM:=$(realpath clapack-wasm)
Vosklet.js: | vosk libarchive
cd $(SRC) && \
em++ -O3 link.cc genericModel.cc recognizer.cc bindings.cc -sWASMFS \ -sWASM_BIGINT -sSINGLE_FILE -sGZIP_EMBEDDINGS -sMODULARIZE -sEMBIND_STD_STRING_IS_UTF8 -sPTHREAD_POOL_DELAY_LOAD -sTEXTDECODER=2 -sPTHREAD_POOL_SIZE_STRICT=2 -sINITIAL_MEMORY=$(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 --pre-js pre.js -o ../Vosklet.js && \
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$(VOSK)/src -l:vosk.a -lembind -pthread -flto -msimd128 -mreference-types -mnontrapping-fptoint -mextended-const -msign-ext --pre-js pre.js -o ../Vosklet.js && \
cd .. && \
rm -f Vosklet.worker.js && \
@@ -68,24 +68,24 @@ openfst: prepare
git clone --depth=1 https://github.com/alphacep/openfst /tmp/openfst && \
cd /tmp/openfst && \
autoreconf -i && \
CXXFLAGS="-pthread -r -O3 -flto -msimd128" LDFLAGS="-O3 -pthread -flto" emconfigure ./configure --prefix=$(OPENFST) --enable-static --disable-shared --enable-ngram-fsts --enable-lookahead-fsts --disable-bin --with-pic && \
CXXFLAGS="-r -O3 -flto -msimd128" LDFLAGS="-O3 -flto" emconfigure ./configure --prefix=$(OPENFST) --enable-static --disable-shared --enable-lookahead-fsts --enable-ngram-fsts --disable-bin && \
emmake make -j$(COMPILE_JOBS) install && \
echo "PACKAGE_VERSION = 1.8.0" >> $(OPENFST)/Makefile && \
rm -rf /tmp/openfst
kaldi: | openfst clapack-wasm
git clone -b vosk --depth=1 https://github.com/alphacep/kaldi $(KALDI) && \
git clone -b vosk --depth=1 https://github.com/alphacep/kaldi && \
cd $(KALDI)/src && \
git apply $(SRC)/kaldi.patch && \
CXXFLAGS="-O3 -UHAVE_EXECINFO_H -pthread -flto -msimd128" LDFLAGS="-O3 -sERROR_ON_UNDEFINED_SYMBOLS=0 -lembind -pthread -flto" emconfigure ./configure --use-cuda=no --with-cudadecoder=no --static --static-math=yes --static-fst=yes --debug-level=0 --double-precision=yes --fst-root=$(OPENFST) --clapack-root=$(CLAPACK_WASM) --host=WASM && \
emmake make -j$(COMPILE_JOBS) online2 lm rnnlm
CXXFLAGS="-O3 -UHAVE_EXECINFO_H -flto -msimd128 -Wno-unused-variable -Wno-unused-but-set-variable -g0" LDFLAGS="-O3 -lembind -flto -g0" emconfigure ./configure --use-cuda=no --with-cudadecoder=no --static --static-math=yes --static-fst=yes --debug-level=0 --fst-root=$(OPENFST) --clapack-root=$(CLAPACK_WASM) --host=WASM && \
emmake make -j$(COMPILE_JOBS) online2 rnnlm
vosk: | kaldi
git clone -b go/v0.3.46 --depth=1 https://github.com/alphacep/vosk-api $(VOSK) && \
cd $(VOSK)/src && \
git apply $(SRC)/vosk.patch && \
VOSK_FILES="recognizer.cc language_model.cc model.cc spk_model.cc vosk_api.cc" && \
em++ -pthread -O3 -flto -msimd128 -Wno-deprecated -I. -I$(KALDI)/src -I$(OPENFST)/include $(VOSK_FILES) -c && \
em++ -O3 -flto -msimd128 -Wno-deprecated -I. -I$(KALDI)/src -I$(OPENFST)/include $(VOSK_FILES) -c && \
emar -rcs vosk.a $(VOSK_FILES:.cc=.o)
.PHONY: prepare

2
src/bindings.cc
View File

@@ -21,7 +21,7 @@ EMSCRIPTEN_BINDINGS() {
.constructor<int, float, genericModel*>(allow_raw_pointers())
.constructor<int, float, genericModel*, genericModel*>(allow_raw_pointers())
.constructor<int, float, genericModel*, std::string, int>(allow_raw_pointers())
.function("acceptWaveform", &recognizer::acceptWaveform, allow_raw_pointers())
.function("pushData", &recognizer::pushData, allow_raw_pointers())
.function("reset", &recognizer::reset, allow_raw_pointers())
.function("setEndpointerMode", &recognizer::setEndpointerMode, allow_raw_pointers())
.function("setEndpointerDelays", &recognizer::setEndpointerDelays, allow_raw_pointers())

11
src/genericModel.cc
View File

@@ -1,10 +1,9 @@
#include "genericModel.h"
genericModel::genericModel(int index, bool normalMdl, std::string storepath, std::string id) : index{index}, normalMdl{normalMdl}, storepath{std::move(storepath)}, id{std::move(id)}, entry{archive_entry_new()} {
blocker.lock();
blocker.acquire();
}
void genericModel::extractAndLoad(int tarStart, int tarSize) {
emscripten_console_log("Untaring");
static fs::path path{};
static int fd{};
func = [this, tarStart, tarSize](){
@@ -12,7 +11,6 @@ void genericModel::extractAndLoad(int tarStart, int tarSize) {
archive_read_support_format_tar(src);
archive_read_open_memory(src, reinterpret_cast<void*>(tarStart), tarSize);
if(archive_errno(src) != 0) {
emscripten_console_logf("Unable to open tar in WASM memory: %s", archive_error_string(src));
free(reinterpret_cast<void*>(tarStart));
fireEv(index, "Unable to open tar in WASM memory");
return;
@@ -32,14 +30,12 @@ void genericModel::extractAndLoad(int tarStart, int tarSize) {
}
fd = creat(path.c_str(),0777);
if(fd == -1) {
emscripten_console_log("Unable to create model files");
fireEv(index, "Unable to create model files");
return;
}
archive_read_data_into_fd(src, fd);
close(fd);
if(archive_errno(src) != 0) {
emscripten_console_logf("Cannot write into model files: %s", archive_error_string(src));
fireEv(index, "Cannot write into model files");
return;
}
@@ -49,14 +45,13 @@ void genericModel::extractAndLoad(int tarStart, int tarSize) {
archive_read_free(src);
if(normalMdl) mdl = vosk_model_new(storepath.c_str());
else vosk_spk_model_new(storepath.c_str());
emscripten_console_log("Model loaded!");
if(normalMdl ? std::get<0>(mdl) == nullptr : std::get<1>(mdl) == nullptr) fireEv(index, "Unable to load model for recognition");
else fireEv(index, "0");
};
std::thread t{[this](){
func();
blocker.lock();
blocker.unlock();
blocker.acquire();
blocker.release();
func();
}};
t.detach();

3
src/genericModel.h
View File

@@ -5,6 +5,7 @@
#include <filesystem>
#include <variant>
#include <thread>
#include <semaphore>
#include <fcntl.h>
#include <emscripten/console.h>
@@ -20,8 +21,8 @@ struct genericModel {
std::string storepath;
std::string id;
std::variant<VoskModel*, VoskSpkModel*> mdl;
std::binary_semaphore blocker{1};
std::function<void()> func;
std::mutex blocker{};
archive_entry* entry;
void extractAndLoad(int tarStart, int tarSize);
genericModel(int index, bool normalMdl, std::string storepath, std::string id);

13
src/kaldi.patch
View File

@@ -39,15 +39,4 @@ index c3a122281..71d37256d 100644
TaskSequencer<IvectorExtractorComputeDerivedVarsClass> sequencer(
sequencer_opts);
for (int32 i = 0; i < NumGauss(); i++)
diff --git a/src/base/kaldi-types.h b/src/base/kaldi-types.h
index 7ebf4f8..2f5979e 100644
--- a/src/base/kaldi-types.h
+++ b/src/base/kaldi-types.h
@@ -20,6 +20,7 @@
#ifndef KALDI_BASE_KALDI_TYPES_H_
#define KALDI_BASE_KALDI_TYPES_H_ 1
+#define KALDI_DOUBLEPRECISION 1
namespace kaldi {
// TYPEDEFS ..................................................................

294
src/pre.js
View File

@@ -1,12 +1,193 @@
let objs = []
let processorURL = URL.createObjectURL(new Blob(['(', (() => {
registerProcessor("VoskletTransferer", class extends AudioWorkletProcessor {
process(inputs) {
this.port.postMessage(inputs[0][0].buffer, [inputs[0][0].buffer])
return true
}
})
}).toString(), ')()'], { type : "text/javascript" }))
let pthreadURL = URL.createObjectURL(new Blob(['(', (() => {
/**
* @license
* Copyright 2015 The Emscripten Authors
* SPDX-License-Identifier: MIT
*/
// Pthread Web Worker startup routine:
// This is the entry point file that is loaded first by each Web Worker
// that executes pthreads on the Emscripten application.
'use strict';
var Module = {};
// Thread-local guard variable for one-time init of the JS state
var initializedJS = false;
function assert(condition, text) {
if (!condition) abort('Assertion failed: ' + text);
}
function threadPrintErr(...args) {
var text = args.join(' ');
console.error(text);
}
function threadAlert(...args) {
var text = args.join(' ');
postMessage({cmd: 'alert', text, threadId: Module['_pthread_self']()});
}
// We don't need out() for now, but may need to add it if we want to use it
// here. Or, if this code all moves into the main JS, that problem will go
// away. (For now, adding it here increases code size for no benefit.)
var out = () => { throw 'out() is not defined in worker.js.'; }
var err = threadPrintErr;
self.alert = threadAlert;
var dbg = threadPrintErr;
Module['instantiateWasm'] = (info, receiveInstance) => {
// Instantiate from the module posted from the main thread.
// We can just use sync instantiation in the worker.
var module = Module['wasmModule'];
// We don't need the module anymore; new threads will be spawned from the main thread.
Module['wasmModule'] = null;
var instance = new WebAssembly.Instance(module, info);
// TODO: Due to Closure regression https://github.com/google/closure-compiler/issues/3193,
// the above line no longer optimizes out down to the following line.
// When the regression is fixed, we can remove this if/else.
return receiveInstance(instance);
}
// Turn unhandled rejected promises into errors so that the main thread will be
// notified about them.
self.onunhandledrejection = (e) => {
throw e.reason || e;
};
function handleMessage(e) {
try {
if (e.data.cmd === 'load') { // Preload command that is called once per worker to parse and load the Emscripten code.
// Until we initialize the runtime, queue up any further incoming messages.
let messageQueue = [];
self.onmessage = (e) => messageQueue.push(e);
// And add a callback for when the runtime is initialized.
self.startWorker = (instance) => {
Module = instance;
// Notify the main thread that this thread has loaded.
postMessage({ 'cmd': 'loaded' });
// Process any messages that were queued before the thread was ready.
for (let msg of messageQueue) {
handleMessage(msg);
}
// Restore the real message handler.
self.onmessage = handleMessage;
};
// Module and memory were sent from main thread
Module['wasmModule'] = e.data.wasmModule;
// Use `const` here to ensure that the variable is scoped only to
// that iteration, allowing safe reference from a closure.
for (const handler of e.data.handlers) {
Module[handler] = (...args) => {
postMessage({ cmd: 'callHandler', handler, args: args });
}
}
Module['wasmMemory'] = e.data.wasmMemory;
Module['buffer'] = Module['wasmMemory'].buffer;
Module['workerID'] = e.data.workerID;
Module['ENVIRONMENT_IS_PTHREAD'] = true;
if (typeof e.data.urlOrBlob == 'string') {
importScripts(e.data.urlOrBlob);
} else {
var objectUrl = URL.createObjectURL(e.data.urlOrBlob);
importScripts(objectUrl);
URL.revokeObjectURL(objectUrl);
}
loadVosklet(Module);
} else if (e.data.cmd === 'run') {
// Pass the thread address to wasm to store it for fast access.
Module['__emscripten_thread_init'](e.data.pthread_ptr, /*is_main=*/0, /*is_runtime=*/0, /*can_block=*/1);
// Await mailbox notifications with `Atomics.waitAsync` so we can start
// using the fast `Atomics.notify` notification path.
Module['__emscripten_thread_mailbox_await'](e.data.pthread_ptr);
assert(e.data.pthread_ptr);
// Also call inside JS module to set up the stack frame for this pthread in JS module scope
Module['establishStackSpace']();
Module['PThread'].receiveObjectTransfer(e.data);
Module['PThread'].threadInitTLS();
if (!initializedJS) {
// Embind must initialize itself on all threads, as it generates support JS.
// We only do this once per worker since they get reused
Module['__embind_initialize_bindings']();
initializedJS = true;
}
try {
Module['invokeEntryPoint'](e.data.start_routine, e.data.arg);
} catch(ex) {
if (ex != 'unwind') {
// The pthread "crashed". Do not call `_emscripten_thread_exit` (which
// would make this thread joinable). Instead, re-throw the exception
// and let the top level handler propagate it back to the main thread.
throw ex;
}
}
} else if (e.data.cmd === 'cancel') { // Main thread is asking for a pthread_cancel() on this thread.
if (Module['_pthread_self']()) {
Module['__emscripten_thread_exit'](-1);
}
} else if (e.data.target === 'setimmediate') {
// no-op
} else if (e.data.cmd === 'checkMailbox') {
if (initializedJS) {
Module['checkMailbox']();
}
} else if (e.data.cmd) {
// The received message looks like something that should be handled by this message
// handler, (since there is a e.data.cmd field present), but is not one of the
// recognized commands:
err(`worker.js received unknown command ${e.data.cmd}`);
err(e.data);
}
} catch(ex) {
err(`worker.js onmessage() captured an uncaught exception: ${ex}`);
if (ex?.stack) err(ex.stack);
Module['__emscripten_thread_crashed']?.();
throw ex;
}
};
self.onmessage = handleMessage;
}).toString(), ')()'], { type : "text/javascript" }))
Module.cleanUp = () => {
objs.forEach(obj => obj.delete())
URL.revokeObjectURL(pthreadUrl)
URL.revokeObjectURL(processorUrl)
objs.forEach(obj => obj.obj.delete())
URL.revokeObjectURL(pthreadURL)
URL.revokeObjectURL(processorURL)
}
Module.createTransferer = async (ctx) => {
await ctx.audioWorklet.addModule(processorURL)
return new AudioWorkletNode(ctx, "VoskletTransferer", {
channelCountMode : "explicit",
numberOfInputs : 1,
numberOfOutputs : 0,
channelCount : 1
})
}
Module.locateFile = (path, scriptDir) => {
if(path === "Vosklet.js") return pthreadUrl
return scriptDir+path
if(path === "Vosklet.worker.js") return pthreadURL
return scriptDir + path
}
async function getFileHandle(path, create = false) {
let components = path.split("/")
@@ -17,16 +198,12 @@ async function getFileHandle(path, create = false) {
return prevDir.getFileHandle(components[components.length - 1], { create : create })
}
class genericModel extends EventTarget {
constructor(url, storepath, id, normalMdl) {
constructor() {
super()
objs.push(this)
this.url = url
this.storepath = storepath
this.id = id
this.normalMdl = normalMdl
}
static async _init(url, storepath, id, normalMdl) {
let mdl = new genericModel(url, storepath, id, normalMdl)
static async create(url, storepath, id, normalMdl) {
let mdl = new genericModel()
let result = new Promise((resolve, reject) => {
mdl.addEventListener("0", ev => {
if(ev.detail === "0") return resolve(mdl)
@@ -37,9 +214,7 @@ class genericModel extends EventTarget {
let tar
mdl.obj = new Module.genericModel(objs.length - 1, normalMdl, "/" + storepath, id)
try {
console.log("Getting Data file")
let dataFile = await (await getFileHandle(storepath + "/model.tgz")).getFile()
console.log("Getting ID file")
let idFile = await (await getFileHandle(storepath + "/id")).getFile()
if(await idFile.text() !== id) throw ""
tar = dataFile.stream()
@@ -71,22 +246,27 @@ class genericModel extends EventTarget {
return result
}
delete() {
if (this.obj) this.obj.delete()
this.obj.delete()
}
}
Module.makeModel = async (url, storepath, id) => {
return genericModel._init(url, storepath, id, true)
Module.createModel = async (url, storepath, id) => {
return genericModel.create(url, storepath, id, true)
}
Module.makeSpkModel = async (url, storepath, id) => {
return genericModel._init(url, storepath, id, false)
Module.createSpkModel = async (url, storepath, id) => {
return genericModel.create(url, storepath, id, false)
}
class Recognizer extends EventTarget {
class recognizer extends EventTarget {
constructor() {
super()
objs.push(this)
return new Proxy(this, {
get(self, prop, receiver) {
return self.obj && Object.keys(Object.getPrototypeOf(self.obj)).includes(prop) ? self.obj[prop].bind(self.obj) : self[prop] ? self[prop].bind ? self[prop].bind(self) : self[prop] : undefined
}
})
}
static async _init(model, sampleRate, mode, grammar, spkModel) {
let rec = new Recognizer()
static async create(model, sampleRate, mode, grammar, spkModel) {
let rec = new recognizer()
let result = new Promise((resolve, reject) => {
rec.addEventListener("0", ev => {
if(ev.detail === "0") return resolve(rec)
@@ -106,66 +286,18 @@ class Recognizer extends EventTarget {
}
return result
}
async getNode(ctx) {
if(typeof this.node === "undefined") {
await ctx.audioWorklet.addModule("../src/processor.js", { credentials : "omit"})
this.node = new AudioWorkletNode(ctx, 'VoskletProcessor', { channelCountMode: "explicit", channelCount: 1, numberOfInputs: 1, numberOfOutputs: 1, processorOptions: { dataBuf: this.dataBuf, state: this.state }})
}
return this.node
}
recognize(buf) {
Module.HEAPF32.set(buf.getChannelData(0).subarray(0, 512), this.ptr)
}
delete() {
if (this.obj) this.obj.delete()
if(this.node) this.node.postMessage(0)
}
setWords(words) {
this.obj.setWords(words)
}
setPartialWords(partialWords) {
this.obj.setPartialWords(partialWords)
}
setGrm(grm) {
this.obj.setGrm(grm)
}
setSpkModel(spkModel) {
this.obj.setSpkModel(spkModel.obj)
}
setNLSML(nlsml) {
this.obj.setNLSML(nlsml)
}
setMaxAlternatives(alts) {
this.obj.setMaxAlternatives(alts)
acceptWaveform(audioData) {
let start = Module._malloc(audioData.length * 4)
Module.HEAPF32.set(audioData, start / 4)
this.obj.pushData(start, audioData.length)
}
}
Module.makeRecognizer = (model, sampleRate) => {
return Recognizer._init(model.obj, sampleRate, 1)
Module.createRecognizer = (model, sampleRate) => {
return recognizer.create(model.obj, sampleRate, 1)
}
Module.makeRecognizerWithSpkModel = (model, sampleRate, spkModel) => {
return Recognizer._init(model.obj, sampleRate, 2, null, spkModel)
Module.createRecognizerWithSpkModel = (model, sampleRate, spkModel) => {
return recognizer.create(model.obj, sampleRate, 2, null, spkModel)
}
Module.makeRecognizerWithGrm = (model, sampleRate, grammar) => {
return Recognizer._init(model.obj, sampleRate, 3, grammar, null)
}
/*let processorURL = URL.createObjectURL(new Blob(['(',
(() => {
registerProcessor("VoskletProcessor", class extends AudioWorkletProcessor {
constructor(options) {
this.channelIndex = options.processorOptions.channelIndex
this.dataBuf = options.processorOptions.dataBuf
this.state = options.processorOptions.state
}
process(inputs, outputs, params) {
while(state[0])
inputs.copyFromChannel(this.dataBuf, 0)
return true
}
})
}).toString(),
')()'], {type : "text/javascript"}))
let pthreadURL = URL.createObjectURL(new Blob(['(',
(() => {
{ PTHREAD_SCRIPT }
}).toString()
, ')()'], {type : "text/javascript"})) */
Module.createRecognizerWithGrm = (model, sampleRate, grammar) => {
return recognizer.create(model.obj, sampleRate, 3, grammar, null)
}

14
src/recognizer.cc
View File

@@ -19,7 +19,10 @@ void recognizer::finishConstruction(genericModel* model, genericModel* spkModel)
return;
}
auto main {[this](){
fireEv(index, "0");
while(!done) {
blocker.acquire();
blocker.release();
while(!dataQ.empty()) {
switch(vosk_recognizer_accept_waveform_f(rec, dataQ.front().data, dataQ.front().len)) {
case 0:
@@ -36,23 +39,22 @@ void recognizer::finishConstruction(genericModel* model, genericModel* spkModel)
if(!model->resourceUsed) {
model->resourceUsed = true;
model->func = main;
model->blocker.unlock();
emscripten_console_log("Using model's thread");
model->blocker.release();
return;
}
if(spkModel != nullptr && !spkModel->resourceUsed) {
spkModel->resourceUsed = true;
spkModel->func = main;
spkModel->blocker.unlock();
emscripten_console_log("Using speaker model's thread");
spkModel->blocker.release();
return;
}
emscripten_console_log("New recognizer thread");
std::thread t{main};
t.detach();
}
void recognizer::acceptWaveform(int start, int len) {
void recognizer::pushData(int start, int len) {
dataQ.emplace(start, len);
blocker.release();
blocker.acquire();
}
void recognizer::reset() {
vosk_recognizer_reset(rec);

8
src/recognizer.h
View File

@@ -1,8 +1,7 @@
#pragma once
#include "genericModel.h"
#include <condition_variable>
#include <queue>
struct audioData {
float* data;
int len;
@@ -10,15 +9,16 @@ struct audioData {
};
struct recognizer {
std::atomic_bool done;
std::queue<audioData> dataQ{};
int index;
std::binary_semaphore blocker{1};
std::queue<audioData> dataQ{};
VoskRecognizer* rec;
recognizer(int index, float sampleRate, genericModel* model);
recognizer(int index, float sampleRate, genericModel* model, genericModel* spkModel);
recognizer(int index, float sampleRate, genericModel* model, const std::string& grm, int dummy);
~recognizer();
void finishConstruction(genericModel* model, genericModel* spkModel = nullptr);
void acceptWaveform(int start, int len);
void pushData(int start, int len);
void reset();
void setEndpointerMode(VoskEndpointerMode mode);
void setEndpointerDelays(float tStartMax, float tEnd, float tMax);