Add WebAssembly for VAD (#1281)

This commit is contained in:
Fangjun Kuang
2024-08-23 17:08:37 +08:00
committed by GitHub
parent fb09f8fae3
commit 1ef8a7a202
20 changed files with 1056 additions and 20 deletions

View File

@@ -10,6 +10,10 @@ if(SHERPA_ONNX_ENABLE_WASM_KWS)
add_subdirectory(kws)
endif()
if(SHERPA_ONNX_ENABLE_WASM_VAD)
add_subdirectory(vad)
endif()
if(SHERPA_ONNX_ENABLE_WASM_NODEJS)
add_subdirectory(nodejs)
endif()

72
wasm/vad/CMakeLists.txt Normal file
View File

@@ -0,0 +1,72 @@
if(NOT $ENV{SHERPA_ONNX_IS_USING_BUILD_WASM_SH})
message(FATAL_ERROR "Please use ./build-wasm-simd-vad.sh to build for wasm VAD")
endif()
if(NOT EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/assets/silero_vad.onnx")
message(FATAL_ERROR "Please read ${CMAKE_CURRENT_SOURCE_DIR}/assets/README.md before you continue")
endif()
set(exported_functions
MyPrint
# VAD
SherpaOnnxCreateCircularBuffer
SherpaOnnxDestroyCircularBuffer
SherpaOnnxCircularBufferPush
SherpaOnnxCircularBufferGet
SherpaOnnxCircularBufferFree
SherpaOnnxCircularBufferPop
SherpaOnnxCircularBufferSize
SherpaOnnxCircularBufferHead
SherpaOnnxCircularBufferReset
SherpaOnnxCreateVoiceActivityDetector
SherpaOnnxDestroyVoiceActivityDetector
SherpaOnnxVoiceActivityDetectorAcceptWaveform
SherpaOnnxVoiceActivityDetectorEmpty
SherpaOnnxVoiceActivityDetectorDetected
SherpaOnnxVoiceActivityDetectorPop
SherpaOnnxVoiceActivityDetectorClear
SherpaOnnxVoiceActivityDetectorFront
SherpaOnnxDestroySpeechSegment
SherpaOnnxVoiceActivityDetectorReset
SherpaOnnxVoiceActivityDetectorFlush
#
)
set(mangled_exported_functions)
foreach(x IN LISTS exported_functions)
list(APPEND mangled_exported_functions "_${x}")
endforeach()
list(JOIN mangled_exported_functions "," all_exported_functions)
include_directories(${CMAKE_SOURCE_DIR})
set(MY_FLAGS " -s FORCE_FILESYSTEM=1 -s INITIAL_MEMORY=64MB -s ALLOW_MEMORY_GROWTH=1")
string(APPEND MY_FLAGS " -sSTACK_SIZE=10485760 ") # 10MB
string(APPEND MY_FLAGS " -sEXPORTED_FUNCTIONS=[_CopyHeap,_malloc,_free,${all_exported_functions}] ")
string(APPEND MY_FLAGS "--preload-file ${CMAKE_CURRENT_SOURCE_DIR}/assets@. ")
string(APPEND MY_FLAGS " -sEXPORTED_RUNTIME_METHODS=['ccall','stringToUTF8','setValue','getValue','lengthBytesUTF8','UTF8ToString'] ")
message(STATUS "MY_FLAGS: ${MY_FLAGS}")
set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} ${MY_FLAGS}")
set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} ${MY_FLAGS}")
set(CMAKE_EXECUTBLE_LINKER_FLAGS "${CMAKE_EXECUTBLE_LINKER_FLAGS} ${MY_FLAGS}")
if (NOT CMAKE_EXECUTABLE_SUFFIX STREQUAL ".js")
message(FATAL_ERROR "The default suffix for building executables should be .js!")
endif()
# set(CMAKE_EXECUTABLE_SUFFIX ".html")
add_executable(sherpa-onnx-wasm-main-vad sherpa-onnx-wasm-main-vad.cc)
target_link_libraries(sherpa-onnx-wasm-main-vad sherpa-onnx-c-api)
install(TARGETS sherpa-onnx-wasm-main-vad DESTINATION bin/wasm/vad)
install(
FILES
"$<TARGET_FILE_DIR:sherpa-onnx-wasm-main-vad>/sherpa-onnx-wasm-main-vad.js"
"index.html"
"sherpa-onnx-vad.js"
"app-vad.js"
"$<TARGET_FILE_DIR:sherpa-onnx-wasm-main-vad>/sherpa-onnx-wasm-main-vad.wasm"
"$<TARGET_FILE_DIR:sherpa-onnx-wasm-main-vad>/sherpa-onnx-wasm-main-vad.data"
DESTINATION
bin/wasm/vad
)

307
wasm/vad/app-vad.js Normal file
View File

@@ -0,0 +1,307 @@
// This file copies and modifies code
// from https://mdn.github.io/web-dictaphone/scripts/app.js
// and https://gist.github.com/meziantou/edb7217fddfbb70e899e
const startBtn = document.getElementById('startBtn');
const stopBtn = document.getElementById('stopBtn');
const clearBtn = document.getElementById('clearBtn');
const hint = document.getElementById('hint');
const soundClips = document.getElementById('sound-clips');
let textArea = document.getElementById('results');
let lastResult = '';
let resultList = [];
clearBtn.onclick = function() {
resultList = [];
textArea.value = getDisplayResult();
textArea.scrollTop = textArea.scrollHeight; // auto scroll
};
function getDisplayResult() {
let i = 0;
let ans = '';
for (let s in resultList) {
if (resultList[s] == '') {
continue;
}
if (resultList[s] == 'Speech detected') {
ans += '' + i + ': ' + resultList[s];
i += 1;
} else {
ans += ', ' + resultList[s] + '\n';
}
}
if (lastResult.length > 0) {
ans += '' + i + ': ' + lastResult + '\n';
}
return ans;
}
Module = {};
Module.onRuntimeInitialized = function() {
console.log('inited!');
hint.innerText = 'Model loaded! Please click start';
startBtn.disabled = false;
vad = createVad(Module);
console.log('vad is created!', vad);
buffer = new CircularBuffer(30 * 16000, Module);
console.log('CircularBuffer is created!', buffer);
};
let audioCtx;
let mediaStream;
let expectedSampleRate = 16000;
let recordSampleRate; // the sampleRate of the microphone
let recorder = null; // the microphone
let leftchannel = []; // TODO: Use a single channel
let recordingLength = 0; // number of samples so far
let vad = null;
let buffer = null;
let printed = false;
if (navigator.mediaDevices.getUserMedia) {
console.log('getUserMedia supported.');
// see https://w3c.github.io/mediacapture-main/#dom-mediadevices-getusermedia
const constraints = {audio: true};
let onSuccess = function(stream) {
if (!audioCtx) {
audioCtx = new AudioContext({sampleRate: expectedSampleRate});
}
console.log(audioCtx);
recordSampleRate = audioCtx.sampleRate;
console.log('sample rate ' + recordSampleRate);
// creates an audio node from the microphone incoming stream
mediaStream = audioCtx.createMediaStreamSource(stream);
console.log('media stream', mediaStream);
// https://developer.mozilla.org/en-US/docs/Web/API/AudioContext/createScriptProcessor
// bufferSize: the onaudioprocess event is called when the buffer is full
var bufferSize = 4096;
var numberOfInputChannels = 1;
var numberOfOutputChannels = 2;
if (audioCtx.createScriptProcessor) {
recorder = audioCtx.createScriptProcessor(
bufferSize, numberOfInputChannels, numberOfOutputChannels);
} else {
recorder = audioCtx.createJavaScriptNode(
bufferSize, numberOfInputChannels, numberOfOutputChannels);
}
console.log('recorder', recorder);
recorder.onaudioprocess = function(e) {
let samples = new Float32Array(e.inputBuffer.getChannelData(0))
samples = downsampleBuffer(samples, expectedSampleRate);
buffer.push(samples);
while (buffer.size() > vad.config.sileroVad.windowSize) {
const s = buffer.get(buffer.head(), vad.config.sileroVad.windowSize);
vad.acceptWaveform(s);
buffer.pop(vad.config.sileroVad.windowSize);
if (vad.isDetected() && !printed) {
printed = true;
lastResult = 'Speech detected';
}
if (!vad.isDetected()) {
printed = false;
if (lastResult != '') {
resultList.push(lastResult);
}
lastResult = '';
}
while (!vad.isEmpty()) {
const segment = vad.front();
const duration = segment.samples.length / expectedSampleRate;
const durationStr = `Duration: ${duration.toFixed(3)} seconds`;
resultList.push(durationStr);
vad.pop();
// now save the segment to a wav file
let buf = new Int16Array(segment.samples.length);
for (var i = 0; i < segment.samples.length; ++i) {
let s = segment.samples[i];
if (s >= 1)
s = 1;
else if (s <= -1)
s = -1;
buf[i] = s * 32767;
}
let clipName = new Date().toISOString() + '--' + durationStr;
const clipContainer = document.createElement('article');
const clipLabel = document.createElement('p');
const audio = document.createElement('audio');
const deleteButton = document.createElement('button');
clipContainer.classList.add('clip');
audio.setAttribute('controls', '');
deleteButton.textContent = 'Delete';
deleteButton.className = 'delete';
clipLabel.textContent = clipName;
clipContainer.appendChild(audio);
clipContainer.appendChild(clipLabel);
clipContainer.appendChild(deleteButton);
soundClips.appendChild(clipContainer);
audio.controls = true;
const blob = toWav(buf);
leftchannel = [];
const audioURL = window.URL.createObjectURL(blob);
audio.src = audioURL;
deleteButton.onclick = function(e) {
let evtTgt = e.target;
evtTgt.parentNode.parentNode.removeChild(evtTgt.parentNode);
};
clipLabel.onclick = function() {
const existingName = clipLabel.textContent;
const newClipName = prompt('Enter a new name for your sound clip?');
if (newClipName === null) {
clipLabel.textContent = existingName;
} else {
clipLabel.textContent = newClipName;
}
};
}
}
textArea.value = getDisplayResult();
textArea.scrollTop = textArea.scrollHeight; // auto scroll
};
startBtn.onclick = function() {
mediaStream.connect(recorder);
recorder.connect(audioCtx.destination);
console.log('recorder started');
stopBtn.disabled = false;
startBtn.disabled = true;
};
stopBtn.onclick = function() {
vad.reset();
buffer.reset();
console.log('recorder stopped');
// stopBtn recording
recorder.disconnect(audioCtx.destination);
mediaStream.disconnect(recorder);
startBtn.style.background = '';
startBtn.style.color = '';
// mediaRecorder.requestData();
stopBtn.disabled = true;
startBtn.disabled = false;
};
};
let onError = function(err) {
console.log('The following error occured: ' + err);
};
navigator.mediaDevices.getUserMedia(constraints).then(onSuccess, onError);
} else {
console.log('getUserMedia not supported on your browser!');
alert('getUserMedia not supported on your browser!');
}
// this function is copied/modified from
// https://gist.github.com/meziantou/edb7217fddfbb70e899e
function flatten(listOfSamples) {
let n = 0;
for (let i = 0; i < listOfSamples.length; ++i) {
n += listOfSamples[i].length;
}
let ans = new Int16Array(n);
let offset = 0;
for (let i = 0; i < listOfSamples.length; ++i) {
ans.set(listOfSamples[i], offset);
offset += listOfSamples[i].length;
}
return ans;
}
// this function is copied/modified from
// https://gist.github.com/meziantou/edb7217fddfbb70e899e
function toWav(samples) {
let buf = new ArrayBuffer(44 + samples.length * 2);
var view = new DataView(buf);
// http://soundfile.sapp.org/doc/WaveFormat/
// F F I R
view.setUint32(0, 0x46464952, true); // chunkID
view.setUint32(4, 36 + samples.length * 2, true); // chunkSize
// E V A W
view.setUint32(8, 0x45564157, true); // format
//
// t m f
view.setUint32(12, 0x20746d66, true); // subchunk1ID
view.setUint32(16, 16, true); // subchunk1Size, 16 for PCM
view.setUint32(20, 1, true); // audioFormat, 1 for PCM
view.setUint16(22, 1, true); // numChannels: 1 channel
view.setUint32(24, expectedSampleRate, true); // sampleRate
view.setUint32(28, expectedSampleRate * 2, true); // byteRate
view.setUint16(32, 2, true); // blockAlign
view.setUint16(34, 16, true); // bitsPerSample
view.setUint32(36, 0x61746164, true); // Subchunk2ID
view.setUint32(40, samples.length * 2, true); // subchunk2Size
let offset = 44;
for (let i = 0; i < samples.length; ++i) {
view.setInt16(offset, samples[i], true);
offset += 2;
}
return new Blob([view], {type: 'audio/wav'});
}
// this function is copied from
// https://github.com/awslabs/aws-lex-browser-audio-capture/blob/master/lib/worker.js#L46
function downsampleBuffer(buffer, exportSampleRate) {
if (exportSampleRate === recordSampleRate) {
return buffer;
}
var sampleRateRatio = recordSampleRate / exportSampleRate;
var newLength = Math.round(buffer.length / sampleRateRatio);
var result = new Float32Array(newLength);
var offsetResult = 0;
var offsetBuffer = 0;
while (offsetResult < result.length) {
var nextOffsetBuffer = Math.round((offsetResult + 1) * sampleRateRatio);
var accum = 0, count = 0;
for (var i = offsetBuffer; i < nextOffsetBuffer && i < buffer.length; i++) {
accum += buffer[i];
count++;
}
result[offsetResult] = accum / count;
offsetResult++;
offsetBuffer = nextOffsetBuffer;
}
return result;
};

View File

@@ -0,0 +1,5 @@
# Introduction
Please download
https://github.com/k2-fsa/sherpa-onnx/releases/download/asr-models/silero_vad.onnx
and put `silero_vad.onnx` into the current directory, i.e., `wasm/vad/assets`.

42
wasm/vad/index.html Normal file
View File

@@ -0,0 +1,42 @@
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width" />
<title>Next-gen Kaldi WebAssembly with sherpa-onnx for Text-to-speech</title>
<style>
h1,div {
text-align: center;
}
textarea {
width:100%;
}
</style>
</head>
<body>
<h1>
Next-gen Kaldi + WebAssembly<br/>
VAD Demo with <a href="https://github.com/k2-fsa/sherpa-onnx">sherpa-onnx</a><br/>
(with <a href="https://github.com/snakers4/silero-vad">silero-vad</a>)
</h1>
<div>
<span id="hint">Loading model ... ...</span>
<br/>
<br/>
<button id="startBtn" disabled>Start</button>
<button id="stopBtn" disabled>Stop</button>
<button id="clearBtn">Clear</button>
<br/>
<br/>
<textarea id="results" rows="10" readonly></textarea>
</div>
<section flex="1" overflow="auto" id="sound-clips">
</section>
<script src="sherpa-onnx-vad.js"></script>
<script src="app-vad.js"></script>
<script src="sherpa-onnx-wasm-main-vad.js"></script>
</body>

253
wasm/vad/sherpa-onnx-vad.js Normal file
View File

@@ -0,0 +1,253 @@
function freeConfig(config, Module) {
if ('buffer' in config) {
Module._free(config.buffer);
}
if ('sileroVad' in config) {
freeConfig(config.sileroVad, Module)
}
Module._free(config.ptr);
}
// The user should free the returned pointers
function initSherpaOnnxSileroVadModelConfig(config, Module) {
const modelLen = Module.lengthBytesUTF8(config.model || '') + 1;
const n = modelLen;
const buffer = Module._malloc(n);
const len = 5 * 4;
const ptr = Module._malloc(len);
Module.stringToUTF8(config.model || '', buffer, modelLen);
offset = 0;
Module.setValue(ptr, buffer, 'i8*');
offset += 4;
Module.setValue(ptr + offset, config.threshold || 0.5, 'float');
offset += 4;
Module.setValue(ptr + offset, config.minSilenceDuration || 0.5, 'float');
offset += 4;
Module.setValue(ptr + offset, config.minSpeechDuration || 0.25, 'float');
offset += 4;
Module.setValue(ptr + offset, config.windowSize || 512, 'i32');
offset += 4;
return {
buffer: buffer, ptr: ptr, len: len,
}
}
function initSherpaOnnxVadModelConfig(config, Module) {
if (!('sileroVad' in config)) {
config.sileroVad = {
model: '',
threshold: 0.50,
minSilenceDuration: 0.50,
minSpeechDuration: 0.25,
windowSize: 512,
};
}
const sileroVad =
initSherpaOnnxSileroVadModelConfig(config.sileroVad, Module);
const len = sileroVad.len + 4 * 4;
const ptr = Module._malloc(len);
const providerLen = Module.lengthBytesUTF8(config.provider || 'cpu') + 1;
const buffer = Module._malloc(providerLen);
Module.stringToUTF8(config.provider || 'cpu', buffer, providerLen);
let offset = 0;
Module._CopyHeap(sileroVad.ptr, sileroVad.len, ptr + offset);
offset += sileroVad.len;
Module.setValue(ptr + offset, config.sampleRate || 16000, 'i32');
offset += 4;
Module.setValue(ptr + offset, config.numThreads || 1, 'i32');
offset += 4;
Module.setValue(ptr + offset, buffer, 'i8*'); // provider
offset += 4;
Module.setValue(ptr + offset, config.debug || 0, 'i32');
offset += 4;
return {
buffer: buffer, ptr: ptr, len: len, sileroVad: sileroVad,
}
}
function createVad(Module, myConfig) {
const sileroVad = {
model: './silero_vad.onnx',
threshold: 0.50,
minSilenceDuration: 0.50,
minSpeechDuration: 0.25,
windowSize: 512,
};
let config = {
sileroVad: sileroVad,
sampleRate: 16000,
numThreads: 1,
provider: 'cpu',
debug: 1,
bufferSizeInSeconds: 30,
};
if (myConfig) {
config = myConfig;
}
return new Vad(config, Module);
}
class CircularBuffer {
constructor(capacity, Module) {
this.handle = Module._SherpaOnnxCreateCircularBuffer(capacity);
this.Module = Module;
}
free() {
this.Module._SherpaOnnxDestroyCircularBuffer(this.handle);
this.handle = 0
}
/**
* @param samples {Float32Array}
*/
push(samples) {
const pointer =
this.Module._malloc(samples.length * samples.BYTES_PER_ELEMENT);
this.Module.HEAPF32.set(samples, pointer / samples.BYTES_PER_ELEMENT);
this.Module._SherpaOnnxCircularBufferPush(
this.handle, pointer, samples.length);
this.Module._free(pointer);
}
get(startIndex, n) {
const p =
this.Module._SherpaOnnxCircularBufferGet(this.handle, startIndex, n);
const samplesPtr = p / 4;
const samples = new Float32Array(n);
for (let i = 0; i < n; i++) {
samples[i] = this.Module.HEAPF32[samplesPtr + i];
}
this.Module._SherpaOnnxCircularBufferFree(p);
return samples;
}
pop(n) {
this.Module._SherpaOnnxCircularBufferPop(this.handle, n);
}
size() {
return this.Module._SherpaOnnxCircularBufferSize(this.handle);
}
head() {
return this.Module._SherpaOnnxCircularBufferHead(this.handle);
}
reset() {
this.Module._SherpaOnnxCircularBufferReset(this.handle);
}
}
class Vad {
constructor(configObj, Module) {
this.config = configObj;
const config = initSherpaOnnxVadModelConfig(configObj, Module);
Module._MyPrint(config.ptr);
const handle = Module._SherpaOnnxCreateVoiceActivityDetector(
config.ptr, configObj.bufferSizeInSeconds || 30);
freeConfig(config, Module);
this.handle = handle;
this.Module = Module;
}
free() {
this.Module._SherpaOnnxDestroyVoiceActivityDetector(this.handle);
this.handle = 0
}
// samples is a float32 array
acceptWaveform(samples) {
const pointer =
this.Module._malloc(samples.length * samples.BYTES_PER_ELEMENT);
this.Module.HEAPF32.set(samples, pointer / samples.BYTES_PER_ELEMENT);
this.Module._SherpaOnnxVoiceActivityDetectorAcceptWaveform(
this.handle, pointer, samples.length);
this.Module._free(pointer);
}
isEmpty() {
return this.Module._SherpaOnnxVoiceActivityDetectorEmpty(this.handle) == 1;
}
isDetected() {
return this.Module._SherpaOnnxVoiceActivityDetectorDetected(this.handle) ==
1;
}
pop() {
this.Module._SherpaOnnxVoiceActivityDetectorPop(this.handle);
}
clear() {
this.Module._SherpaOnnxVoiceActivityDetectorClear(this.handle);
}
/*
{
samples: a 1-d float32 array,
start: an int32
}
*/
front() {
const h = this.Module._SherpaOnnxVoiceActivityDetectorFront(this.handle);
const start = this.Module.HEAP32[h / 4];
const samplesPtr = this.Module.HEAP32[h / 4 + 1] / 4;
const numSamples = this.Module.HEAP32[h / 4 + 2];
const samples = new Float32Array(numSamples);
for (let i = 0; i < numSamples; i++) {
samples[i] = this.Module.HEAPF32[samplesPtr + i];
}
this.Module._SherpaOnnxDestroySpeechSegment(h);
return {samples: samples, start: start};
}
reset() {
this.Module._SherpaOnnxVoiceActivityDetectorReset(this.handle);
}
flush() {
this.Module._SherpaOnnxVoiceActivityDetectorFlush(this.handle);
}
};
if (typeof process == 'object' && typeof process.versions == 'object' &&
typeof process.versions.node == 'string') {
module.exports = {
createVad,
CircularBuffer,
};
}

View File

@@ -0,0 +1,45 @@
// wasm/sherpa-onnx-wasm-main-vad.cc
//
// Copyright (c) 2024 Xiaomi Corporation
#include <stdio.h>
#include <algorithm>
#include <memory>
#include "sherpa-onnx/c-api/c-api.h"
// see also
// https://emscripten.org/docs/porting/connecting_cpp_and_javascript/Interacting-with-code.html
extern "C" {
static_assert(sizeof(SherpaOnnxSileroVadModelConfig) == 5 * 4, "");
static_assert(sizeof(SherpaOnnxVadModelConfig) ==
sizeof(SherpaOnnxSileroVadModelConfig) + 4 * 4,
"");
void MyPrint(SherpaOnnxVadModelConfig *config) {
auto silero_vad = &config->silero_vad;
fprintf(stdout, "----------silero_vad config----------\n");
fprintf(stdout, "model: %s\n", silero_vad->model);
fprintf(stdout, "threshold: %.3f\n", silero_vad->threshold);
fprintf(stdout, "min_silence_duration: %.3f\n",
silero_vad->min_silence_duration);
fprintf(stdout, "min_speech_duration: %.3f\n",
silero_vad->min_speech_duration);
fprintf(stdout, "window_size: %d\n", silero_vad->window_size);
fprintf(stdout, "----------config----------\n");
fprintf(stdout, "sample_rate: %d\n", config->sample_rate);
fprintf(stdout, "num_threads: %d\n", config->num_threads);
fprintf(stdout, "provider: %s\n", config->provider);
fprintf(stdout, "debug: %d\n", config->debug);
}
void CopyHeap(const char *src, int32_t num_bytes, char *dst) {
std::copy(src, src + num_bytes, dst);
}
}