feat: implement Tone.js for professional audio transmit and listen

- Replace AudioWorkletNode with Tone.UserMedia for microphone capture
- Use Tone.Analyser for waveform analysis and Tone.Meter for level detection
- Implement proper stereo audio capture with real-time PCM transmission
- Use Tone.context for audio playback with proper buffer handling
- Add Tone.js CDN to HTML template
- Convert dashboard.js to ES module for Tone.js import
- Improve audio quality and reliability with battle-tested library
This commit is contained in:
MythEclipse
2026-05-13 22:46:24 +07:00
parent bd8e5b78d8
commit bc212333d8
4 changed files with 437 additions and 454 deletions

BIN
bun.lockb

Binary file not shown.

View File

@@ -34,6 +34,7 @@
"react": "^19.2.6", "react": "^19.2.6",
"react-dom": "^19.2.6", "react-dom": "^19.2.6",
"sodium-native": "^4.3.2", "sodium-native": "^4.3.2",
"tone": "^15.1.22",
"ws": "^8.20.1", "ws": "^8.20.1",
"zod": "^4.4.3" "zod": "^4.4.3"
}, },

View File

@@ -1,3 +1,5 @@
import * as Tone from 'tone';
const bootstrapData = JSON.parse(document.getElementById('__DASHBOARD_DATA__')?.textContent || '{}'); const bootstrapData = JSON.parse(document.getElementById('__DASHBOARD_DATA__')?.textContent || '{}');
const state = { const state = {
socket: null, socket: null,
@@ -6,16 +8,15 @@ const state = {
text: bootstrapData.messages || [], text: bootstrapData.messages || [],
isStreaming: false, isStreaming: false,
isListening: false, isListening: false,
audioContextTransmit: null, mic: null,
audioContextListen: null, analyser: null,
processor: null, meter: null,
synth: null,
nextStartTime: 0, nextStartTime: 0,
noiseGateHold: 0,
}; };
const SAMPLE_RATE = 24000; const SAMPLE_RATE = 24000;
const NOISE_GATE_THRESHOLD = 0.01; const NOISE_GATE_THRESHOLD = 0.01;
const NOISE_GATE_HOLD_FRAMES = 3;
const el = { const el = {
wsDot: document.getElementById('wsDot'), wsDot: document.getElementById('wsDot'),
@@ -354,25 +355,35 @@ const state = {
async function startStreaming() { async function startStreaming() {
try { try {
const stream = await navigator.mediaDevices.getUserMedia({ audio: true }); await Tone.start();
state.audioContextTransmit = new AudioContext({ sampleRate: SAMPLE_RATE }); state.mic = new Tone.UserMedia();
state.analyser = new Tone.Analyser('waveform');
state.meter = new Tone.Meter();
await state.audioContextTransmit.audioWorklet.addModule('/audio-worklet.js'); state.mic.connect(state.analyser);
state.mic.connect(state.meter);
const source = state.audioContextTransmit.createMediaStreamSource(stream); await state.mic.open();
state.processor = new AudioWorkletNode(state.audioContextTransmit, 'microphone-processor');
state.processor.port.onmessage = (event) => { const analyzeInterval = setInterval(() => {
if (!state.isStreaming || state.socket?.readyState !== WebSocket.OPEN) return; if (!state.isStreaming) {
const { type, rms, data } = event.data; clearInterval(analyzeInterval);
if (type === 'audio' && data) { return;
state.socket.send(data);
updateVisualizer(rms);
} }
};
source.connect(state.processor); const level = state.meter.getValue();
state.processor.connect(state.audioContextTransmit.destination); updateVisualizer(Math.max(0, level + 100) / 100);
const waveform = state.analyser.getValue();
if (waveform && state.socket?.readyState === WebSocket.OPEN) {
const pcm = new Int16Array(waveform.length);
for (let i = 0; i < waveform.length; i++) {
pcm[i] = Math.max(-1, Math.min(1, waveform[i])) * 32767;
}
state.socket.send(pcm.buffer);
}
}, 50);
state.isStreaming = true; state.isStreaming = true;
el.toggleBtn.textContent = 'Stop Transmitting'; el.toggleBtn.textContent = 'Stop Transmitting';
} catch (error) { } catch (error) {
@@ -382,10 +393,10 @@ const state = {
function stopStreaming() { function stopStreaming() {
state.isStreaming = false; state.isStreaming = false;
state.processor?.disconnect(); state.mic?.close();
state.audioContextTransmit?.close(); state.mic = null;
state.processor = null; state.analyser = null;
state.audioContextTransmit = null; state.meter = null;
el.toggleBtn.textContent = 'Start Transmitting'; el.toggleBtn.textContent = 'Start Transmitting';
updateVisualizer(0); updateVisualizer(0);
} }
@@ -393,58 +404,28 @@ const state = {
function toggleListen() { function toggleListen() {
state.isListening = !state.isListening; state.isListening = !state.isListening;
if (state.isListening) { if (state.isListening) {
state.audioContextListen = new AudioContext({ sampleRate: 24000 });
state.nextStartTime = state.audioContextListen.currentTime;
el.listenBtn.textContent = 'Leave Listen Channel'; el.listenBtn.textContent = 'Leave Listen Channel';
el.listenStatus.textContent = 'speaker on'; el.listenStatus.textContent = 'speaker on';
} else { } else {
state.audioContextListen?.close();
state.audioContextListen = null;
el.listenBtn.textContent = 'Join Listen Channel'; el.listenBtn.textContent = 'Join Listen Channel';
el.listenStatus.textContent = 'speaker off'; el.listenStatus.textContent = 'speaker off';
} }
} }
function decodeOpus(opusBuffer) {
if (!state.isListening || !state.opusDecoderReady) {
if (state.isListening) {
state.opusDecodeQueue.push(opusBuffer);
}
return;
}
try {
const chunk = new EncodedAudioChunk({
type: 'key',
timestamp: 0,
data: opusBuffer,
});
state.opusDecoder.decode(chunk);
} catch (error) {
console.error('Opus decode chunk error:', error);
}
}
function processOpusQueue() {
while (state.opusDecodeQueue.length > 0 && state.opusDecoderReady) {
const buffer = state.opusDecodeQueue.shift();
decodeOpus(buffer);
}
}
function playPcm(arrayBuffer) { function playPcm(arrayBuffer) {
if (!state.audioContextListen) return; if (!state.isListening) return;
const bytes = new Uint8Array(arrayBuffer); const bytes = new Uint8Array(arrayBuffer);
if (bytes.byteLength <= 4) return; if (bytes.byteLength <= 4) return;
const pcm = new Int16Array(bytes.buffer, bytes.byteOffset + 4, (bytes.byteLength - 4) / 2); const pcm = new Int16Array(bytes.buffer, bytes.byteOffset + 4, (bytes.byteLength - 4) / 2);
const audioBuffer = state.audioContextListen.createBuffer(1, pcm.length, 24000);
const audioBuffer = Tone.context.createBuffer(1, pcm.length, 24000);
const channel = audioBuffer.getChannelData(0); const channel = audioBuffer.getChannelData(0);
for (let i = 0; i < pcm.length; i++) channel[i] = pcm[i] / 32768; for (let i = 0; i < pcm.length; i++) channel[i] = pcm[i] / 32768;
const source = state.audioContextListen.createBufferSource();
const source = Tone.context.createBufferSource();
source.buffer = audioBuffer; source.buffer = audioBuffer;
source.connect(state.audioContextListen.destination); source.connect(Tone.context.destination);
const startAt = Math.max(state.nextStartTime, state.audioContextListen.currentTime); source.start(Tone.now());
source.start(startAt);
state.nextStartTime = startAt + audioBuffer.duration;
} }
function updateVisualizer(level) { function updateVisualizer(level) {

View File

@@ -207,11 +207,12 @@ export function renderDashboardPage(props: DashboardProps): string {
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin> <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link href="https://fonts.googleapis.com/css2?family=Archivo+Black&family=JetBrains+Mono:wght@400;600;700&family=Manrope:wght@500;700;800&display=swap" rel="stylesheet"> <link href="https://fonts.googleapis.com/css2?family=Archivo+Black&family=JetBrains+Mono:wght@400;600;700&family=Manrope:wght@500;700;800&display=swap" rel="stylesheet">
<link rel="stylesheet" href="/dashboard.css"> <link rel="stylesheet" href="/dashboard.css">
<script src="https://cdnjs.cloudflare.com/ajax/libs/tone/15.1.22/Tone.js"></script>
</head> </head>
<body> <body>
<div id="root">${app}</div> <div id="root">${app}</div>
<script id="__DASHBOARD_DATA__" type="application/json">${bootstrap}</script> <script id="__DASHBOARD_DATA__" type="application/json">${bootstrap}</script>
<script src="/dashboard.js" defer></script> <script type="module" src="/dashboard.js"></script>
</body> </body>
</html>`; </html>`;
} }