Spaces:
Running
Running
// This file copies and modifies code | |
// from https://mdn.github.io/web-dictaphone/scripts/app.js | |
// and https://gist.github.com/meziantou/edb7217fddfbb70e899e | |
const urlParams = new URLSearchParams(window.location.search); | |
const lang = urlParams.get('lang') ?? 'en'; | |
const activeTab = urlParams.get('tab') ?? 'single'; | |
const startBtn = document.getElementById('recordBtn'); | |
const hint = document.getElementById('hint'); | |
const soundClips = document.getElementById('fileInput'); | |
const playAllBtn = document.getElementById('playAllBtn'); | |
let started = false; | |
let multistreamStarted = false; | |
let textArea = document.getElementById('results'); | |
let lastResult = ''; | |
let resultList = []; | |
function clear() { | |
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; | |
} | |
ans += '' + i + ': ' + resultList[s] + '\n'; | |
i += 1; | |
} | |
if (lastResult.length > 0) { | |
ans += '' + i + ': ' + lastResult + '\n'; | |
} | |
return ans; | |
} | |
Module = {}; | |
Module.locateFile = function(path, scriptDirectory = '') { | |
if (path.endsWith('.js.metadata')) { | |
return scriptDirectory + path.replace('.js.metadata', '.json'); | |
} | |
return scriptDirectory + path; | |
}; | |
Module.setStatus = function(status) { | |
const statusElement = document.getElementById('status'); | |
statusElement.textContent = status; | |
if (status === '') { | |
statusElement.style.display = 'none'; | |
document.querySelectorAll('.tab-content').forEach((tabContentElement) => { | |
tabContentElement.classList.remove('loading'); | |
}); | |
} else { | |
statusElement.style.display = 'block'; | |
document.querySelectorAll('.tab-content').forEach((tabContentElement) => { | |
tabContentElement.classList.add('loading'); | |
}); | |
} | |
}; | |
Module.onRuntimeInitialized = function() { | |
console.log('inited!'); | |
//hint.innerText = 'Model loaded! Please click start'; | |
started = false; | |
recognizer = createOnlineRecognizer(Module); | |
console.log('recognizer is created!', recognizer); | |
}; | |
function loadScript(src) { | |
const scriptElement = document.createElement('script'); | |
scriptElement.src = src; | |
document.body.append(scriptElement); | |
} | |
loadScript('./' + lang + '.js'); | |
loadScript('./sherpa-onnx-wasm-main-asr.js'); | |
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 recognizer = null; | |
let recognizer_stream = null; | |
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: 16000}); | |
} | |
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); | |
if (recognizer_stream == null) { | |
recognizer_stream = recognizer.createStream(); | |
} | |
recognizer_stream.acceptWaveform(expectedSampleRate, samples); | |
while (recognizer.isReady(recognizer_stream)) { | |
recognizer.decode(recognizer_stream); | |
} | |
let isEndpoint = recognizer.isEndpoint(recognizer_stream); | |
let result = recognizer.getResult(recognizer_stream).text; | |
if (recognizer.config.modelConfig.paraformer.encoder != '') { | |
let tailPaddings = new Float32Array(expectedSampleRate); | |
recognizer_stream.acceptWaveform(expectedSampleRate, tailPaddings); | |
while (recognizer.isReady(recognizer_stream)) { | |
recognizer.decode(recognizer_stream); | |
} | |
result = recognizer.getResult(recognizer_stream).text; | |
} | |
if (result.length > 0 && lastResult != result) { | |
lastResult = result; | |
} | |
if (isEndpoint) { | |
if (lastResult.length > 0) { | |
resultList.push(lastResult); | |
lastResult = ''; | |
} | |
recognizer.reset(recognizer_stream); | |
} | |
textArea.value = getDisplayResult(); | |
textArea.scrollTop = textArea.scrollHeight; // auto scroll | |
let buf = new Int16Array(samples.length); | |
for (var i = 0; i < samples.length; ++i) { | |
let s = samples[i]; | |
if (s >= 1) | |
s = 1; | |
else if (s <= -1) | |
s = -1; | |
samples[i] = s; | |
buf[i] = s * 32767; | |
} | |
leftchannel.push(buf); | |
recordingLength += bufferSize; | |
}; | |
startBtn.onclick = function() { | |
if(started) { | |
console.log('recorder stopped'); | |
recorder.disconnect(audioCtx.destination); | |
mediaStream.disconnect(recorder); | |
started = false; | |
var clipName = new Date().toISOString(); | |
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; | |
let samples = flatten(leftchannel); | |
const blob = toWav(samples); | |
leftchannel = []; | |
const audioURL = window.URL.createObjectURL(blob); | |
audio.src = audioURL; | |
console.log('recorder stopped'); | |
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; | |
} | |
}; | |
} | |
else { | |
mediaStream.connect(recorder); | |
recorder.connect(audioCtx.destination); | |
console.log('recorder started'); | |
started = true; | |
} | |
}; | |
}; | |
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!'); | |
} | |
playAllBtn.onclick = function() { | |
if(!multistreamStarted) { | |
multistreamStarted = true; | |
playAllBtn.textContent = "Stop All Streams"; | |
playAllBtn.style.backgroundColor = "#d9534f"; | |
transcribe(audioSources[lang][0], 'transcript1', 0); | |
transcribe(audioSources[lang][1], 'transcript2', 1); | |
transcribe(audioSources[lang][2], 'transcript3', 2); | |
transcribe(audioSources[lang][3], 'transcript4', 3); | |
transcribe(audioSources[lang][4], 'transcript5', 4); | |
} | |
else { | |
audios[0].pause(); | |
audios[1].pause(); | |
audios[2].pause(); | |
audios[3].pause(); | |
audios[4].pause(); | |
audios[0].currentTime = 0; | |
audios[1].currentTime = 0; | |
audios[2].currentTime = 0; | |
audios[3].currentTime = 0; | |
audios[4].currentTime = 0; | |
playAllBtn.textContent = "Play All Streams"; | |
playAllBtn.style.backgroundColor = "#007bff"; | |
multistreamStarted = false; | |
} | |
} | |
// 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; | |
}; | |
async function processArrayBufferWithASR(arrayBuffer, file) { | |
// Check if recognizer is ready. | |
if (recognizer === null) { | |
console.error("Recognizer not yet initialized! Please wait for WASM to load."); | |
//resultsTextarea.value = "Error: Recognizer not ready."; | |
return; | |
} | |
// Create an AudioContext. (On some platforms, creating multiple AudioContexts can be problematic. | |
// If needed, consider reusing a global AudioContext.) | |
const audioCtx = new (window.AudioContext || window.webkitAudioContext)({ | |
sampleRate: expectedSampleRate | |
}); | |
try { | |
const audioBuffer = await audioCtx.decodeAudioData(arrayBuffer); | |
console.log("AudioBuffer decoded. Duration (s):", audioBuffer.duration); | |
let channelData = audioBuffer.getChannelData(0); | |
console.log("Channel data length:", channelData.length); | |
// Downsample if necessary. | |
if (audioBuffer.sampleRate !== expectedSampleRate) { | |
console.log("Downsampling from", audioBuffer.sampleRate, "to", expectedSampleRate); | |
channelData = downsampleBuffer(channelData, expectedSampleRate, audioBuffer.sampleRate); | |
console.log("Downsampled channel data length:", channelData.length); | |
} | |
// Create a new recognizer stream. | |
const stream = recognizer.createStream(); | |
const chunkSize = expectedSampleRate; // assume 1 second worth of samples per chunk. | |
for (let i = 0; i < channelData.length; i += chunkSize) { | |
const chunk = channelData.subarray(i, i + chunkSize); | |
stream.acceptWaveform(expectedSampleRate, chunk); | |
while (recognizer.isReady(stream)) { | |
recognizer.decode(stream); | |
} | |
} | |
// Flush any tail data if necessary. | |
const tail = new Float32Array(expectedSampleRate); | |
stream.acceptWaveform(expectedSampleRate, tail); | |
while (recognizer.isReady(stream)) { | |
recognizer.decode(stream); | |
} | |
const fileResult = recognizer.getResult(stream).text || ""; | |
console.log("ASR result for file:", fileResult); | |
textArea.value = fileResult; | |
} catch (err) { | |
console.error("Error decoding audio data:", err); | |
//resultsTextarea.value = "Error processing audio: " + err.message; | |
} | |
} | |
const audios = [] | |
const recorders = [] | |
async function loadAudio(url) { | |
try { | |
const response = await fetch(url, { mode: "cors" }); | |
if (!response.ok) throw new Error("Network response was not ok"); | |
const blob = await response.blob(); | |
const objectUrl = URL.createObjectURL(blob); | |
return new Audio(objectUrl); | |
} catch (error) { | |
console.error("Error loading audio:", error); | |
} | |
} | |
async function transcribe(url, output, index) { | |
let urlLabel = document.createElement('h2'); | |
//urlLabel.textContent = url; | |
let textarea = document.getElementById(output); | |
textarea.value = ''; | |
//textarea.readOnly = true; | |
//textarea.rows = 10; | |
//document.querySelector('#container').append(urlLabel); | |
//document.querySelector('#container').append(textarea); | |
let lastResult = ''; | |
let resultList = []; | |
function getDisplayResult() { | |
let i = 0; | |
let ans = ''; | |
for (let s in resultList) { | |
if (resultList[s] == '') { | |
continue; | |
} | |
ans += '' + i + ': ' + resultList[s] + '\n'; | |
i += 1; | |
} | |
if (lastResult.length > 0) { | |
ans += '' + i + ': ' + lastResult + '\n'; | |
} | |
return ans; | |
} | |
if(!audios[index]) { | |
audios[index] = await loadAudio(url); | |
} | |
audios[index].play(); | |
console.log(audioCtx); | |
let recordSampleRate = audioCtx.sampleRate; | |
console.log('sample rate ' + recordSampleRate); | |
// 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(!recorders[index]) { | |
recorders[index] = audioCtx.createScriptProcessor( | |
bufferSize, numberOfInputChannels, numberOfOutputChannels); | |
let mediaStream = audioCtx.createMediaElementSource(audios[index]); | |
mediaStream.connect(recorders[index]); | |
recorders[index].connect(audioCtx.destination); | |
} | |
let recognizer_stream = null; | |
recorders[index].onaudioprocess = function(e) { | |
let samples = new Float32Array(e.inputBuffer.getChannelData(0)) | |
e.outputBuffer.copyToChannel(samples, 0); | |
if (recognizer_stream == null) { | |
recognizer_stream = recognizer.createStream(); | |
} | |
recognizer_stream.acceptWaveform(expectedSampleRate, samples); | |
while (recognizer.isReady(recognizer_stream)) { | |
recognizer.decode(recognizer_stream); | |
} | |
let isEndpoint = recognizer.isEndpoint(recognizer_stream); | |
let result = recognizer.getResult(recognizer_stream).text; | |
if (result.length > 0 && lastResult != result) { | |
lastResult = result; | |
} | |
if (isEndpoint) { | |
if (lastResult.length > 0) { | |
resultList.push(lastResult); | |
lastResult = ''; | |
} | |
recognizer.reset(recognizer_stream); | |
} | |
textarea.value = getDisplayResult(); | |
textarea.scrollTop = textarea.scrollHeight; // auto scroll | |
let buf = new Int16Array(samples.length); | |
for (var i = 0; i < samples.length; ++i) { | |
let s = samples[i]; | |
if (s >= 1) | |
s = 1; | |
else if (s <= -1) | |
s = -1; | |
samples[i] = s; | |
buf[i] = s * 32767; | |
} | |
}; | |
recorders[index]?.addEventListener("recordingStopped", () => { | |
console.log("Decoding has stopped."); | |
mediaStream.disconnect(recorders[index]); | |
}); | |
} | |
soundClips.addEventListener("change", function (event) { | |
if (!event.target.files || !event.target.files[0]) { | |
console.log("No file selected."); | |
return; | |
} | |
const file = event.target.files[0]; | |
console.log("Selected file:", file.name, file.type, file.size, "bytes"); | |
const reader = new FileReader(); | |
reader.onload = function (ev) { | |
console.log("FileReader onload called."); | |
const arrayBuffer = ev.target.result; | |
console.log("ArrayBuffer length:", arrayBuffer.byteLength); | |
var url = URL.createObjectURL(file); | |
transcribe(url, 'results'); | |
//processArrayBufferWithASR(arrayBuffer, file); | |
}; | |
reader.onerror = function (err) { | |
console.error("FileReader error:", err); | |
}; | |
console.log("Starting FileReader.readAsArrayBuffer..."); | |
reader.readAsArrayBuffer(file); | |
}); | |
const singleAudioTab = document.getElementById("singleAudioTab"); | |
const multistreamTab = document.getElementById("multistreamTab"); | |
const singleAudioContent = document.getElementById("singleAudioContent"); | |
const multistreamContent = document.getElementById("multistreamContent"); | |
const audioElements = [ | |
document.getElementById("audio1"), | |
document.getElementById("audio2"), | |
document.getElementById("audio3"), | |
document.getElementById("audio4"), | |
document.getElementById("audio5"), | |
]; | |
const audioSources = { | |
"de": [ | |
"./de1.mp3", | |
"./de2.mp3", | |
"./de3.mp3", | |
"./de4.mp3", | |
"./de5.mp3", | |
], | |
"en": [ | |
"./en1.mp3", | |
"./en2.mp3", | |
"./en3.mp3", | |
"./en4.mp3", | |
"./en5.mp3", | |
], | |
"fr": [ | |
"./fr1.mp3", | |
"./fr2.mp3", | |
"./fr3.mp3", | |
"./fr4.mp3", | |
"./fr5.mp3", | |
], | |
}; | |
// Tab switching logic | |
singleAudioTab.addEventListener("click", () => { | |
singleAudioTab.classList.add("active"); | |
multistreamTab.classList.remove("active"); | |
singleAudioContent.style.display = "block"; | |
multistreamContent.style.display = "none"; | |
singleAudioTab.style.borderBottomColor = "#007bff"; | |
multistreamTab.style.borderBottomColor = "transparent"; | |
singleAudioTab.style.color = "#007bff"; | |
multistreamTab.style.color = "#6c757d"; | |
const url = new URL(window.location.href); | |
url.searchParams.set("tab", "single"); | |
window.history.pushState({}, "", url.toString()); | |
}); | |
multistreamTab.addEventListener("click", () => { | |
multistreamTab.classList.add("active"); | |
singleAudioTab.classList.remove("active"); | |
multistreamContent.style.display = "block"; | |
singleAudioContent.style.display = "none"; | |
multistreamTab.style.borderBottomColor = "#007bff"; | |
singleAudioTab.style.borderBottomColor = "transparent"; | |
multistreamTab.style.color = "#007bff"; | |
singleAudioTab.style.color = "#6c757d"; | |
const url = new URL(window.location.href); | |
url.searchParams.set("tab", "multi"); | |
window.history.pushState({}, "", url.toString()); | |
}); | |
// Load audio sources | |
audioElements.forEach((audio, index) => { | |
audio.src = audioSources[lang][index]; | |
}); | |
// Microphone recording logic | |
const recordBtn = document.getElementById("recordBtn"); | |
const outputText = document.getElementById("outputText"); | |
const audioPlayback = document.getElementById("audioPlayback"); | |
let mediaRecorder; | |
let audioChunks = []; | |
recordBtn.addEventListener("click", async () => { | |
if (!mediaRecorder || mediaRecorder.state === "inactive") { | |
try { | |
const stream = await navigator.mediaDevices.getUserMedia({ audio: true }); | |
mediaRecorder = new MediaRecorder(stream); | |
mediaRecorder.ondataavailable = (event) => { | |
audioChunks.push(event.data); | |
}; | |
mediaRecorder.onstop = () => { | |
const audioBlob = new Blob(audioChunks, { type: "audio/wav" }); | |
audioChunks = []; | |
const audioURL = URL.createObjectURL(audioBlob); | |
audioPlayback.src = audioURL; | |
audioPlayback.style.display = "block"; | |
//outputText.value = "Recording completed. Playback is ready."; | |
}; | |
mediaRecorder.start(); | |
recordBtn.textContent = "Stop Recording"; | |
recordBtn.style.color = "#5cb85c"; | |
//outputText.value = "Recording..."; | |
} catch (err) { | |
//outputText.value = "Error accessing microphone: " + err.message; | |
} | |
} else if (mediaRecorder.state === "recording") { | |
mediaRecorder.stop(); | |
recordBtn.textContent = "Use Microphone"; | |
recordBtn.style.color = "#d9534f"; | |
} | |
}); | |
// Function to handle language change and update URL | |
document.querySelectorAll('input[name="language"]').forEach((input) => { | |
input.addEventListener('change', function() { | |
const selectedLang = this.value; | |
// Update URL with new language parameter | |
const url = new URL(window.location.href); | |
url.searchParams.set('lang', selectedLang); | |
// Reload the page with new language setting | |
window.location.href = url.toString(); | |
}); | |
}); | |
document.querySelectorAll('input[name="language"]').forEach((input) => { | |
input.checked = input.value === lang; | |
}); | |
if (activeTab === "multi") { | |
multistreamTab.classList.add("active"); | |
singleAudioTab.classList.remove("active"); | |
multistreamContent.style.display = "block"; | |
singleAudioContent.style.display = "none"; | |
multistreamTab.style.borderBottomColor = "#007bff"; | |
singleAudioTab.style.borderBottomColor = "transparent"; | |
multistreamTab.style.color = "#007bff"; | |
singleAudioTab.style.color = "#6c757d"; | |
} |