mirror of
https://github.com/HumanAIGC-Engineering/gradio-webrtc.git
synced 2026-02-05 18:09:23 +08:00
837330dcd8
* Turn integration * Add code: * type hint * Fix typehint * add code * format * WIP * trickle ice * bump version * Better docs * Modify * code * Mute icon for whisper * Add code * llama 4 demo * code * OpenAI interruptions * fix docs
565 lines
19 KiB
HTML
565 lines
19 KiB
HTML
<!DOCTYPE html>
|
|
<html lang="en">
|
|
|
|
<head>
|
|
<meta charset="UTF-8">
|
|
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
|
<title>Real-time Whisper Transcription</title>
|
|
<style>
|
|
:root {
|
|
--primary-gradient: linear-gradient(135deg, #f9a45c 0%, #e66465 100%);
|
|
--background-cream: #faf8f5;
|
|
--background-cream-end: #f7f5f2;
|
|
/* Slightly warmer end color for body gradient */
|
|
--text-dark: #2d2d2d;
|
|
--transcript-bg: #ffffff;
|
|
/* White background for transcript area */
|
|
--transcript-border: #e0e0e0;
|
|
/* Light border for transcript items */
|
|
}
|
|
|
|
body {
|
|
font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen, Ubuntu, Cantarell, sans-serif;
|
|
margin: 0;
|
|
padding: 0;
|
|
/* Apply a subtle vertical gradient to the body */
|
|
background: linear-gradient(to bottom, var(--background-cream), var(--background-cream-end));
|
|
color: var(--text-dark);
|
|
min-height: 100vh;
|
|
}
|
|
|
|
.hero {
|
|
background: var(--primary-gradient);
|
|
color: white;
|
|
padding: 2.5rem 2rem;
|
|
text-align: center;
|
|
}
|
|
|
|
.hero h1 {
|
|
font-size: 2.5rem;
|
|
margin: 0;
|
|
font-weight: 600;
|
|
letter-spacing: -0.5px;
|
|
}
|
|
|
|
.hero p {
|
|
font-size: 1rem;
|
|
margin-top: 0.5rem;
|
|
opacity: 0.9;
|
|
}
|
|
|
|
.container {
|
|
max-width: 1000px;
|
|
margin: 2.5rem auto;
|
|
/* Increased top/bottom margin */
|
|
padding: 0 2rem;
|
|
}
|
|
|
|
.transcript-container {
|
|
border-radius: 12px;
|
|
/* Slightly larger radius */
|
|
box-shadow: 0 4px 15px rgba(0, 0, 0, 0.08);
|
|
/* Enhanced shadow */
|
|
padding: 1.5rem;
|
|
height: 350px;
|
|
/* Increased height */
|
|
overflow-y: auto;
|
|
margin-bottom: 2rem;
|
|
/* Increased margin */
|
|
border: 1px solid rgba(0, 0, 0, 0.05);
|
|
/* Softer border */
|
|
background-color: var(--transcript-bg);
|
|
/* Use the new variable */
|
|
}
|
|
|
|
.controls {
|
|
text-align: center;
|
|
margin: 1.5rem 0;
|
|
}
|
|
|
|
button {
|
|
background: var(--primary-gradient);
|
|
color: white;
|
|
border: none;
|
|
padding: 10px 20px;
|
|
font-size: 0.95rem;
|
|
border-radius: 6px;
|
|
cursor: pointer;
|
|
transition: all 0.2s ease;
|
|
font-weight: 500;
|
|
min-width: 180px;
|
|
position: relative;
|
|
padding-right: 50px;
|
|
}
|
|
|
|
button:hover {
|
|
transform: translateY(-1px);
|
|
box-shadow: 0 4px 12px rgba(230, 100, 101, 0.15);
|
|
}
|
|
|
|
button:active {
|
|
transform: translateY(0);
|
|
}
|
|
|
|
/* Transcript text styling */
|
|
.transcript-container p {
|
|
margin: 0.6rem 0;
|
|
/* Increased vertical margin */
|
|
padding: 0.8rem 1rem;
|
|
/* Increased padding */
|
|
background: var(--background-cream);
|
|
/* Use the lighter cream for contrast */
|
|
border-radius: 6px;
|
|
/* Slightly larger radius */
|
|
line-height: 1.5;
|
|
/* Improved line spacing */
|
|
font-size: 0.98rem;
|
|
/* Slightly larger font */
|
|
border-left: 3px solid var(--transcript-border);
|
|
/* Add a subtle left border */
|
|
transition: background-color 0.2s ease;
|
|
/* Smooth hover effect */
|
|
}
|
|
|
|
.transcript-container p:hover {
|
|
background-color: #fdfbf9;
|
|
/* Slightly change background on hover */
|
|
}
|
|
|
|
/* Custom scrollbar - update track color */
|
|
.transcript-container::-webkit-scrollbar {
|
|
width: 8px;
|
|
/* Slightly wider scrollbar */
|
|
}
|
|
|
|
.transcript-container::-webkit-scrollbar-track {
|
|
background: var(--background-cream-end);
|
|
/* Match body end gradient */
|
|
border-radius: 4px;
|
|
}
|
|
|
|
.transcript-container::-webkit-scrollbar-thumb {
|
|
background: #e66465;
|
|
border-radius: 3px;
|
|
opacity: 0.8;
|
|
}
|
|
|
|
.transcript-container::-webkit-scrollbar-thumb:hover {
|
|
background: #f9a45c;
|
|
}
|
|
|
|
/* Add styles for toast notifications */
|
|
.toast {
|
|
position: fixed;
|
|
top: 20px;
|
|
left: 50%;
|
|
transform: translateX(-50%);
|
|
padding: 16px 24px;
|
|
border-radius: 4px;
|
|
font-size: 14px;
|
|
z-index: 1000;
|
|
display: none;
|
|
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.2);
|
|
}
|
|
|
|
.toast.error {
|
|
background-color: #f44336;
|
|
color: white;
|
|
}
|
|
|
|
.toast.warning {
|
|
background-color: #ffd700;
|
|
color: black;
|
|
}
|
|
|
|
/* Add styles for audio visualization */
|
|
.icon-with-spinner {
|
|
display: flex;
|
|
align-items: center;
|
|
justify-content: center;
|
|
gap: 12px;
|
|
min-width: 180px;
|
|
}
|
|
|
|
.spinner {
|
|
width: 20px;
|
|
height: 20px;
|
|
border: 2px solid white;
|
|
border-top-color: transparent;
|
|
border-radius: 50%;
|
|
animation: spin 1s linear infinite;
|
|
flex-shrink: 0;
|
|
}
|
|
|
|
.pulse-container {
|
|
display: flex;
|
|
align-items: center;
|
|
justify-content: center;
|
|
gap: 12px;
|
|
min-width: 180px;
|
|
}
|
|
|
|
.pulse-circle {
|
|
width: 20px;
|
|
height: 20px;
|
|
border-radius: 50%;
|
|
background-color: white;
|
|
opacity: 0.2;
|
|
flex-shrink: 0;
|
|
transform: translateX(-0%) scale(var(--audio-level, 1));
|
|
transition: transform 0.1s ease;
|
|
}
|
|
|
|
/* Styles for the mute button */
|
|
.mute-toggle {
|
|
position: absolute;
|
|
right: 10px;
|
|
top: 50%;
|
|
transform: translateY(-50%);
|
|
width: 24px;
|
|
height: 24px;
|
|
cursor: pointer;
|
|
display: flex;
|
|
align-items: center;
|
|
justify-content: center;
|
|
}
|
|
|
|
.mute-toggle svg {
|
|
width: 20px;
|
|
height: 20px;
|
|
stroke: white;
|
|
}
|
|
|
|
/* Adjust layout for button content when mute is present */
|
|
.button-content {
|
|
display: flex;
|
|
align-items: center;
|
|
justify-content: center;
|
|
width: calc(100% - 40px);
|
|
margin-right: 40px;
|
|
}
|
|
|
|
.icon-with-spinner,
|
|
.pulse-container {
|
|
width: 100%;
|
|
}
|
|
|
|
@keyframes spin {
|
|
to {
|
|
transform: rotate(360deg);
|
|
}
|
|
}
|
|
</style>
|
|
</head>
|
|
|
|
<body>
|
|
<!-- Add toast element after body opening tag -->
|
|
<div id="error-toast" class="toast"></div>
|
|
<div class="hero">
|
|
<h1>Real-time Transcription</h1>
|
|
<p>Powered by Groq and FastRTC</p>
|
|
</div>
|
|
|
|
<div class="container">
|
|
<div class="transcript-container" id="transcript">
|
|
</div>
|
|
<div class="controls">
|
|
<button id="start-button">Start Recording</button>
|
|
</div>
|
|
</div>
|
|
|
|
<script>
|
|
let peerConnection;
|
|
let webrtc_id;
|
|
let audioContext, analyser, audioSource;
|
|
let audioLevel = 0;
|
|
let animationFrame;
|
|
let isMuted = false;
|
|
|
|
const startButton = document.getElementById('start-button');
|
|
const transcriptDiv = document.getElementById('transcript');
|
|
|
|
// SVG Icons
|
|
const micIconSVG = `
|
|
<svg xmlns="http://www.w3.org/2000/svg" width="100%" height="100%" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
|
|
<path d="M12 1a3 3 0 0 0-3 3v8a3 3 0 0 0 6 0V4a3 3 0 0 0-3-3z"></path>
|
|
<path d="M19 10v2a7 7 0 0 1-14 0v-2"></path>
|
|
<line x1="12" y1="19" x2="12" y2="23"></line>
|
|
<line x1="8" y1="23" x2="16" y2="23"></line>
|
|
</svg>`;
|
|
|
|
const micMutedIconSVG = `
|
|
<svg xmlns="http://www.w3.org/2000/svg" width="100%" height="100%" viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
|
|
<path d="M12 1a3 3 0 0 0-3 3v8a3 3 0 0 0 6 0V4a3 3 0 0 0-3-3z"></path>
|
|
<path d="M19 10v2a7 7 0 0 1-14 0v-2"></path>
|
|
<line x1="12" y1="19" x2="12" y2="23"></line>
|
|
<line x1="8" y1="23" x2="16" y2="23"></line>
|
|
<line x1="1" y1="1" x2="23" y2="23"></line>
|
|
</svg>`;
|
|
|
|
function showError(message) {
|
|
const toast = document.getElementById('error-toast');
|
|
toast.textContent = message;
|
|
toast.style.display = 'block';
|
|
|
|
// Hide toast after 5 seconds
|
|
setTimeout(() => {
|
|
toast.style.display = 'none';
|
|
}, 5000);
|
|
}
|
|
|
|
async function handleMessage(event) {
|
|
// Handle any WebRTC data channel messages if needed
|
|
const eventJson = JSON.parse(event.data);
|
|
if (eventJson.type === "error") {
|
|
showError(eventJson.message);
|
|
} else if (eventJson.type === "send_input") {
|
|
const response = await fetch('/send_input', {
|
|
method: 'POST',
|
|
headers: { 'Content-Type': 'application/json' },
|
|
body: JSON.stringify({
|
|
webrtc_id: webrtc_id,
|
|
transcript: ""
|
|
})
|
|
});
|
|
}
|
|
console.log('Received message:', event.data);
|
|
|
|
}
|
|
|
|
function updateButtonState() {
|
|
// Remove existing mute listener if present
|
|
const existingMuteButton = startButton.querySelector('.mute-toggle');
|
|
if (existingMuteButton) {
|
|
existingMuteButton.removeEventListener('click', toggleMute);
|
|
existingMuteButton.remove();
|
|
}
|
|
|
|
if (peerConnection && (peerConnection.connectionState === 'connecting' || peerConnection.connectionState === 'new')) {
|
|
startButton.innerHTML = `
|
|
<div class="button-content">
|
|
<div class="icon-with-spinner">
|
|
<div class="spinner"></div>
|
|
<span>Connecting...</span>
|
|
</div>
|
|
</div>
|
|
`;
|
|
startButton.disabled = true;
|
|
} else if (peerConnection && peerConnection.connectionState === 'connected') {
|
|
startButton.innerHTML = `
|
|
<div class="button-content">
|
|
<div class="pulse-container">
|
|
<div class="pulse-circle"></div>
|
|
<span>Stop Recording</span>
|
|
</div>
|
|
</div>
|
|
<div class="mute-toggle" title="${isMuted ? 'Unmute' : 'Mute'}">
|
|
${isMuted ? micMutedIconSVG : micIconSVG}
|
|
</div>
|
|
`;
|
|
startButton.disabled = false;
|
|
const muteButton = startButton.querySelector('.mute-toggle');
|
|
if (muteButton) {
|
|
muteButton.addEventListener('click', toggleMute);
|
|
}
|
|
} else {
|
|
startButton.innerHTML = 'Start Recording';
|
|
startButton.disabled = false;
|
|
}
|
|
}
|
|
|
|
function toggleMute(event) {
|
|
event.stopPropagation();
|
|
if (!peerConnection || peerConnection.connectionState !== 'connected') return;
|
|
|
|
isMuted = !isMuted;
|
|
console.log("Mute toggled:", isMuted);
|
|
|
|
peerConnection.getSenders().forEach(sender => {
|
|
if (sender.track && sender.track.kind === 'audio') {
|
|
sender.track.enabled = !isMuted;
|
|
console.log(`Audio track ${sender.track.id} enabled: ${!isMuted}`);
|
|
}
|
|
});
|
|
|
|
updateButtonState();
|
|
}
|
|
|
|
function setupAudioVisualization(stream) {
|
|
audioContext = new (window.AudioContext || window.webkitAudioContext)();
|
|
analyser = audioContext.createAnalyser();
|
|
audioSource = audioContext.createMediaStreamSource(stream);
|
|
audioSource.connect(analyser);
|
|
analyser.fftSize = 64;
|
|
const dataArray = new Uint8Array(analyser.frequencyBinCount);
|
|
|
|
function updateAudioLevel() {
|
|
analyser.getByteFrequencyData(dataArray);
|
|
const average = Array.from(dataArray).reduce((a, b) => a + b, 0) / dataArray.length;
|
|
audioLevel = average / 255;
|
|
|
|
const pulseCircle = document.querySelector('.pulse-circle');
|
|
if (pulseCircle) {
|
|
pulseCircle.style.setProperty('--audio-level', 1 + audioLevel);
|
|
}
|
|
|
|
animationFrame = requestAnimationFrame(updateAudioLevel);
|
|
}
|
|
updateAudioLevel();
|
|
}
|
|
|
|
async function setupWebRTC() {
|
|
const config = __RTC_CONFIGURATION__;
|
|
peerConnection = new RTCPeerConnection(config);
|
|
|
|
const timeoutId = setTimeout(() => {
|
|
const toast = document.getElementById('error-toast');
|
|
toast.textContent = "Connection is taking longer than usual. Are you on a VPN?";
|
|
toast.className = 'toast warning';
|
|
toast.style.display = 'block';
|
|
|
|
// Hide warning after 5 seconds
|
|
setTimeout(() => {
|
|
toast.style.display = 'none';
|
|
}, 5000);
|
|
}, 5000);
|
|
|
|
try {
|
|
const stream = await navigator.mediaDevices.getUserMedia({
|
|
audio: true
|
|
});
|
|
|
|
setupAudioVisualization(stream);
|
|
|
|
stream.getTracks().forEach(track => {
|
|
peerConnection.addTrack(track, stream);
|
|
});
|
|
|
|
// Add connection state change listener
|
|
peerConnection.addEventListener('connectionstatechange', () => {
|
|
console.log('connectionstatechange', peerConnection.connectionState);
|
|
if (peerConnection.connectionState === 'connected') {
|
|
clearTimeout(timeoutId);
|
|
const toast = document.getElementById('error-toast');
|
|
toast.style.display = 'none';
|
|
}
|
|
updateButtonState();
|
|
});
|
|
|
|
peerConnection.onicecandidate = ({ candidate }) => {
|
|
if (candidate) {
|
|
console.debug("Sending ICE candidate", candidate);
|
|
fetch('/webrtc/offer', {
|
|
method: 'POST',
|
|
headers: { 'Content-Type': 'application/json' },
|
|
body: JSON.stringify({
|
|
candidate: candidate.toJSON(),
|
|
webrtc_id: webrtc_id,
|
|
type: "ice-candidate",
|
|
})
|
|
})
|
|
}
|
|
};
|
|
|
|
// Create data channel for messages
|
|
const dataChannel = peerConnection.createDataChannel('text');
|
|
dataChannel.onmessage = handleMessage;
|
|
|
|
// Create and send offer
|
|
const offer = await peerConnection.createOffer();
|
|
await peerConnection.setLocalDescription(offer);
|
|
|
|
webrtc_id = Math.random().toString(36).substring(7);
|
|
|
|
const response = await fetch('/webrtc/offer', {
|
|
method: 'POST',
|
|
headers: { 'Content-Type': 'application/json' },
|
|
body: JSON.stringify({
|
|
sdp: peerConnection.localDescription.sdp,
|
|
type: peerConnection.localDescription.type,
|
|
webrtc_id: webrtc_id
|
|
})
|
|
});
|
|
|
|
const serverResponse = await response.json();
|
|
|
|
if (serverResponse.status === 'failed') {
|
|
showError(serverResponse.meta.error === 'concurrency_limit_reached'
|
|
? `Too many connections. Maximum limit is ${serverResponse.meta.limit}`
|
|
: serverResponse.meta.error);
|
|
stop();
|
|
startButton.textContent = 'Start Recording';
|
|
return;
|
|
}
|
|
|
|
await peerConnection.setRemoteDescription(serverResponse);
|
|
|
|
// Create event stream to receive transcripts
|
|
const eventSource = new EventSource('/transcript?webrtc_id=' + webrtc_id);
|
|
eventSource.addEventListener("output", (event) => {
|
|
appendTranscript(event.data);
|
|
});
|
|
} catch (err) {
|
|
clearTimeout(timeoutId);
|
|
console.error('Error setting up WebRTC:', err);
|
|
showError('Failed to establish connection. Please try again.');
|
|
stop();
|
|
startButton.textContent = 'Start Recording';
|
|
}
|
|
}
|
|
|
|
function appendTranscript(text) {
|
|
const p = document.createElement('p');
|
|
p.textContent = text;
|
|
transcriptDiv.appendChild(p);
|
|
transcriptDiv.scrollTop = transcriptDiv.scrollHeight;
|
|
}
|
|
|
|
function stop() {
|
|
if (animationFrame) {
|
|
cancelAnimationFrame(animationFrame);
|
|
animationFrame = null;
|
|
}
|
|
if (audioContext) {
|
|
audioContext.close().catch(e => console.error("Error closing AudioContext:", e));
|
|
audioContext = null;
|
|
analyser = null;
|
|
audioSource = null;
|
|
}
|
|
if (peerConnection) {
|
|
if (peerConnection.getSenders) {
|
|
peerConnection.getSenders().forEach(sender => {
|
|
if (sender.track) {
|
|
sender.track.stop();
|
|
console.log(`Track ${sender.track.id} stopped.`);
|
|
}
|
|
});
|
|
}
|
|
peerConnection.close();
|
|
peerConnection = null;
|
|
console.log("Peer connection closed.");
|
|
}
|
|
audioLevel = 0;
|
|
isMuted = false;
|
|
updateButtonState();
|
|
}
|
|
|
|
startButton.addEventListener('click', (event) => {
|
|
if (event.target.closest('.mute-toggle')) {
|
|
return;
|
|
}
|
|
|
|
if (peerConnection && peerConnection.connectionState === 'connected') {
|
|
console.log("Stop button clicked");
|
|
stop();
|
|
} else if (!peerConnection || ['new', 'closed', 'failed', 'disconnected'].includes(peerConnection.connectionState)) {
|
|
console.log("Start button clicked");
|
|
transcriptDiv.innerHTML = '';
|
|
setupWebRTC();
|
|
updateButtonState();
|
|
}
|
|
});
|
|
</script>
|
|
</body>
|
|
|
|
</html> |