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

Revamp event system to not switch from C++ --> JS twice. Fix transferer node offset

Browse Source
This commit is contained in:
msqr1
2024-10-02 20:42:39 -07:00
parent da3c7a269a
commit 30228eb0c1
9 changed files with 49 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
Examples/Vosklet.js
View File

File diff suppressed because one or more lines are too long

BIN
Examples/Vosklet.wasm
View File

Binary file not shown.

2
Vosklet.js
View File

File diff suppressed because one or more lines are too long

BIN
Vosklet.wasm
View File

Binary file not shown.

32
src/Bindings.cc
View File

@@ -5,7 +5,7 @@
using namespace emscripten; using namespace emscripten;
EMSCRIPTEN_BINDINGS() { EMSCRIPTEN_BINDINGS() {
function("setLogLevel", &vosk_set_log_level, allow_raw_pointers()); function("setLogLevel", &vosk_set_log_level);
enum_<VoskEndpointerMode>("EpMode") enum_<VoskEndpointerMode>("EpMode")
.value("ANSWER_DEFAULT", VOSK_EP_ANSWER_DEFAULT) .value("ANSWER_DEFAULT", VOSK_EP_ANSWER_DEFAULT)
@@ -14,22 +14,22 @@ EMSCRIPTEN_BINDINGS() {
.value("ANSWER_VERY_LONG", VOSK_EP_ANSWER_VERY_LONG); .value("ANSWER_VERY_LONG", VOSK_EP_ANSWER_VERY_LONG);
class_<CommonModel>("CommonModel") class_<CommonModel>("CommonModel")
.constructor<int, bool, std::string, std::string, int, int>(allow_raw_pointers()) .constructor<int, bool, std::string, std::string, int, int>(return_value_policy::take_ownership())
.function("findWord", &CommonModel::findWord, allow_raw_pointers()); .function("findWord", &CommonModel::findWord);
class_<Recognizer>("Recognizer") class_<Recognizer>("Recognizer")
.constructor<int, float, CommonModel*>(allow_raw_pointers()) .constructor<int, float, CommonModel*>(return_value_policy::take_ownership())
.constructor<int, float, CommonModel*, CommonModel*>(allow_raw_pointers()) .constructor<int, float, CommonModel*, CommonModel*>(return_value_policy::take_ownership())
.constructor<int, float, CommonModel*, std::string, int>(allow_raw_pointers()) .constructor<int, float, CommonModel*, std::string, int>(return_value_policy::take_ownership())
.function("safeDelete", &Recognizer::safeDelete, allow_raw_pointers()) .function("safeDelete", &Recognizer::safeDelete)
.function("acceptWaveform", &Recognizer::acceptWaveform, allow_raw_pointers()) .function("acceptWaveform", &Recognizer::acceptWaveform)
.function("reset", &Recognizer::reset, allow_raw_pointers()) .function("reset", &Recognizer::reset)
.function("setEndpointerMode", &Recognizer::setEndpointerMode, allow_raw_pointers()) .function("setEndpointerMode", &Recognizer::setEndpointerMode)
.function("setEndpointerDelays", &Recognizer::setEndpointerDelays, allow_raw_pointers()) .function("setEndpointerDelays", &Recognizer::setEndpointerDelays)
.function("setWords", &Recognizer::setWords, allow_raw_pointers()) .function("setWords", &Recognizer::setWords)
.function("setPartialWords", &Recognizer::setPartialWords, allow_raw_pointers()) .function("setPartialWords", &Recognizer::setPartialWords)
.function("setGrm", &Recognizer::setGrm, allow_raw_pointers()) .function("setGrm", &Recognizer::setGrm)
.function("setNLSML", &Recognizer::setNLSML, allow_raw_pointers()) .function("setNLSML", &Recognizer::setNLSML)
.function("setSpkModel", &Recognizer::setSpkModel, allow_raw_pointers()) .function("setSpkModel", &Recognizer::setSpkModel, allow_raw_pointers())
.function("setMaxAlternatives", &Recognizer::setMaxAlternatives, allow_raw_pointers()); .function("setMaxAlternatives", &Recognizer::setMaxAlternatives);
}; };

8
src/Recognizer.cc
View File

@@ -1,20 +1,20 @@
#include "Recognizer.h" #include "Recognizer.h"
#include <atomic> #include <atomic>
static const char* recognizerInitErr{"Unable to initialize recognizer"}; const char* recognizerInitErr{"Unable to initialize recognizer"};
Recognizer::Recognizer(int index, float sampleRate, CommonModel* model) : Recognizer::Recognizer(int index, float sampleRate, CommonModel* model) :
rec{vosk_recognizer_new(std::get<VoskModel*>(model->mdl), sampleRate)} { rec{vosk_recognizer_new(std::get<VoskModel*>(model->mdl), sampleRate)} {
if(rec == nullptr) mtFireEv(index, Event::status, reinterpret_cast<int>(recognizerInitErr)); if(rec == nullptr) fireEv(index, Event::status, recognizerInitErr);
else globalPool.exec([this, index]{main(index);}); else globalPool.exec([this, index]{main(index);});
} }
Recognizer::Recognizer(int index, float sampleRate, CommonModel* model, CommonModel* spkModel) : Recognizer::Recognizer(int index, float sampleRate, CommonModel* model, CommonModel* spkModel) :
rec{vosk_recognizer_new_spk(std::get<VoskModel*>(model->mdl), sampleRate, std::get<VoskSpkModel*>(spkModel->mdl))} { rec{vosk_recognizer_new_spk(std::get<VoskModel*>(model->mdl), sampleRate, std::get<VoskSpkModel*>(spkModel->mdl))} {
if(rec == nullptr) mtFireEv(index, Event::status, reinterpret_cast<int>(recognizerInitErr)); if(rec == nullptr) fireEv(index, Event::status, recognizerInitErr);
else globalPool.exec([this, index]{main(index);}); else globalPool.exec([this, index]{main(index);});
} }
Recognizer::Recognizer(int index, float sampleRate, CommonModel* model, const std::string& grm, int) : Recognizer::Recognizer(int index, float sampleRate, CommonModel* model, const std::string& grm, int) :
rec{vosk_recognizer_new_grm(std::get<VoskModel*>(model->mdl), sampleRate, grm.c_str())} { rec{vosk_recognizer_new_grm(std::get<VoskModel*>(model->mdl), sampleRate, grm.c_str())} {
if(rec == nullptr) mtFireEv(index, Event::status, reinterpret_cast<int>(recognizerInitErr)); if(rec == nullptr) fireEv(index, Event::status, recognizerInitErr);
else globalPool.exec([this, index]{main(index);}); else globalPool.exec([this, index]{main(index);});
} }
void Recognizer::safeDelete(bool _processCurrent) { void Recognizer::safeDelete(bool _processCurrent) {

25
src/Util.cc
View File

@@ -3,12 +3,12 @@
#include "emscripten/wasm_worker.h" #include "emscripten/wasm_worker.h"
#include "emscripten/em_js.h" #include "emscripten/em_js.h"
EM_JS(void, mtFireEv, (int index, int typeIdx, int content), { EM_JS(void, fireEv, (int idx, int typeIdx, const char* content), {
objs[index].dispatchEvent(new CustomEvent(events[typeIdx], { "detail" : content == 0 ? null : UTF8ToString(content) })); if(ENVIRONMENT_IS_WEB) objs[idx].dispatchEvent(new CustomEvent(events[typeIdx], {
"detail": content == 0 ? null : UTF8ToString(content)
}));
else self.postMessage([idx, typeIdx, content]);
}); });
void fireEv(int index, int typeIdx, const char* content) {
emscripten_wasm_worker_post_function_viii(0, mtFireEv, index, typeIdx, reinterpret_cast<int>(content));
}
int untar(unsigned char* tar, int tarSize, const std::string& storepath) { int untar(unsigned char* tar, int tarSize, const std::string& storepath) {
if(std::memcmp(tar + 257, "ustar", 5)) return IncorrectFormat; if(std::memcmp(tar + 257, "ustar", 5)) return IncorrectFormat;
size_t size{}; size_t size{};
@@ -49,7 +49,7 @@ int untar(unsigned char* tar, int tarSize, const std::string& storepath) {
} }
return Successful; return Successful;
} }
static void workerStartup(int _pool) { void workerStartup(int _pool) {
WorkerPool& pool{*reinterpret_cast<WorkerPool*>(_pool)}; WorkerPool& pool{*reinterpret_cast<WorkerPool*>(_pool)};
std::function<void()> fn; std::function<void()> fn;
while(!pool.done) { while(!pool.done) {
@@ -73,15 +73,16 @@ static void workerStartup(int _pool) {
} }
using _startupFn = void(*)(int); using _startupFn = void(*)(int);
EM_JS(void, startupWorkers, (_startupFn startupFn, WorkerPool* pool), { EM_JS(void, startupWorkers, (_startupFn startupFn, WorkerPool* pool), {
Object.values(_wasmWorkers).forEach(worker => { for(let worker of Object.values(_wasmWorkers)) {
worker.postMessage({ worker.postMessage({
"_wsc": startupFn, "_wsc": startupFn,
"x": [ pool ] "x": [ pool ]
}); });
worker.onmessage = msg => fireEv(...msg.data);
}
}) })
}) constexpr int workerStack{32768};
static constexpr int workerStack{32768}; std::array<std::byte, MAX_WORKERS * workerStack> stacks;
static std::array<std::byte, MAX_WORKERS * workerStack> stacks;
WorkerPool::WorkerPool() { WorkerPool::WorkerPool() {
for(int i{}; i < MAX_WORKERS; ++i) { for(int i{}; i < MAX_WORKERS; ++i) {
emscripten_create_wasm_worker(&stacks[i * workerStack], workerStack); emscripten_create_wasm_worker(&stacks[i * workerStack], workerStack);
@@ -90,10 +91,14 @@ WorkerPool::WorkerPool() {
} }
#undef MAX_WORKERS #undef MAX_WORKERS
WorkerPool::~WorkerPool() { WorkerPool::~WorkerPool() {
// LTO will remove the EM_JS definition for some reason if it isn't called in the same translation unit (I get undefined symbols), even though it is annotated with EMSCRIPTEN_KEEPALIVE. "Call" it here (this destructor is never called) to workaround that. I'm going to file an issue on Emscripten
fireEv(0, 0);
/*
done = true; done = true;
emscripten_atomic_store_u32(&qLock, false); emscripten_atomic_store_u32(&qLock, false);
emscripten_atomic_notify(&qLock, -1); emscripten_atomic_notify(&qLock, -1);
emscripten_terminate_all_wasm_workers(); emscripten_terminate_all_wasm_workers();
*/
} }
void WorkerPool::exec(std::function<void()> fn) { void WorkerPool::exec(std::function<void()> fn) {
taskQ.emplace(fn); taskQ.emplace(fn);

6
src/Util.h
View File

@@ -41,11 +41,7 @@ struct WorkerPool {
~WorkerPool(); ~WorkerPool();
void exec(std::function<void()> fn); void exec(std::function<void()> fn);
}; };
// Must be called from the main thread extern "C" void fireEv(int idx, int typeIdx, const char* content = nullptr);
extern "C" void mtFireEv(int index, int typeIdx, int content);
// Must be called from a wasm worker
void fireEv(int index, int typeIdx, const char* content = nullptr);
int untar(unsigned char* tar, int tarSize, const std::string& storepath); int untar(unsigned char* tar, int tarSize, const std::string& storepath);

21
src/Wrapper.js
View File

@@ -51,17 +51,18 @@ let processorURL = URL.createObjectURL(new Blob(['(', (() => {
registerProcessor('VoskletTransferer', class extends AudioWorkletProcessor { registerProcessor('VoskletTransferer', class extends AudioWorkletProcessor {
constructor(opts) { constructor(opts) {
super(); super();
this.filledSize = 0; this.filled = 0;
this.bufferSize = opts.processorOptions.bufferSize; this.bufSize = opts.processorOptions[0];
this.buffer = new Float32Array(this.bufferSize); this.buf = new Float32Array(this.bufSize);
} }
process(inputs) { process(inputs) {
if(inputs[0][0]) { if(inputs[0][0]) {
this.buffer.set(inputs[0][0], this.filledSize += 128); this.buf.set(inputs[0][0], this.filled);
if(this.filledSize >= this.bufferSize) { this.filled += 128;
this.filledSize = 0; if(this.filled >= this.bufSize) {
this.port.postMessage(this.buffer, [this.buffer.buffer]); this.filled = 0;
this.buffer = new Float32Array(this.bufferSize); this.port.postMessage(this.buf, [this.buf.buffer]);
this.buf = new Float32Array(this.bufSize);
} }
} }
return true; return true;
@@ -158,14 +159,14 @@ Module = {
storageWorker.terminate(); storageWorker.terminate();
}, },
'createTransferer': async (ctx, bufferSize) => { 'createTransferer': async (ctx, bufSize) => {
await ctx.audioWorklet.addModule(processorURL); await ctx.audioWorklet.addModule(processorURL);
return new AudioWorkletNode(ctx, 'VoskletTransferer', { return new AudioWorkletNode(ctx, 'VoskletTransferer', {
channelCountMode: 'explicit', channelCountMode: 'explicit',
numberOfInputs: 1, numberOfInputs: 1,
numberOfOutputs: 0, numberOfOutputs: 0,
channelCount: 1, channelCount: 1,
processorOptions: { bufferSize: bufferSize } processorOptions: [bufSize]
}); });
}, },