Files
2026-06-20 17:38:19 +05:00

484 lines
16 KiB
JavaScript

const TARGET_SAMPLE_RATE = 16000;
const startBtn = document.getElementById("startBtn");
const stopBtn = document.getElementById("stopBtn");
const resetBtn = document.getElementById("resetBtn");
const statusEl = document.getElementById("status");
const partialLine = document.getElementById("partialLine");
const messagesEl = document.getElementById("messages");
const canvas = document.getElementById("waveform");
const canvasCtx = canvas.getContext("2d");
const levelEl = document.getElementById("level");
const muteLabel = document.getElementById("muteLabel");
let socket;
let mediaStream;
let audioContext;
let sourceNode;
let processorNode;
let isRunning = false;
let isStopping = false;
let muteMic = false;
let currentAssistantLine = null;
let currentAssistantText = "";
let currentAssistantAudioChunks = 0;
let currentAssistantTtsFailed = false;
let currentAssistantTtsMessage = "";
let lastErrorMessage = "";
let playbackCursor = 0;
let pendingUnmuteTimer = null;
let encodedPlaybackChain = Promise.resolve();
let pendingEncodedAudioCount = 0;
const activeEncodedAudios = new Set();
drawIdleWave();
startBtn.addEventListener("click", startCall);
stopBtn.addEventListener("click", stopCall);
resetBtn.addEventListener("click", resetSession);
async function startCall() {
lastErrorMessage = "";
isStopping = false;
startBtn.disabled = true;
setStatus("connecting", "busy");
try {
socket = new WebSocket(`${location.protocol === "https:" ? "wss" : "ws"}://${location.host}/ws`);
socket.binaryType = "arraybuffer";
socket.addEventListener("message", handleServerMessage);
socket.addEventListener("close", () => {
if (isRunning) stopCall(lastErrorMessage || "Звонок остановлен.");
if (!statusEl.classList.contains("error")) setStatus("offline", "");
});
socket.addEventListener("error", () => {
lastErrorMessage = "Ошибка WebSocket.";
setStatus("socket error", "error");
partialLine.textContent = lastErrorMessage;
addMessage("system", lastErrorMessage);
});
await waitForSocketOpen(socket);
await waitForServerReady(socket);
mediaStream = await navigator.mediaDevices.getUserMedia({
audio: {
echoCancellation: true,
noiseSuppression: true,
autoGainControl: true,
channelCount: 1,
},
});
audioContext = new AudioContext();
await audioContext.resume();
sourceNode = audioContext.createMediaStreamSource(mediaStream);
processorNode = audioContext.createScriptProcessor(4096, 1, 1);
processorNode.onaudioprocess = handleAudioProcess;
sourceNode.connect(processorNode);
processorNode.connect(audioContext.destination);
isRunning = true;
stopBtn.disabled = false;
resetBtn.disabled = false;
setStatus("live", "live");
partialLine.textContent = "Слушаю...";
socket.send(JSON.stringify({ type: "start_greeting" }));
} catch (error) {
lastErrorMessage = error.message || "Не удалось начать звонок.";
await cleanupAfterFailedStart();
startBtn.disabled = false;
stopBtn.disabled = true;
resetBtn.disabled = true;
setStatus("error", "error");
partialLine.textContent = lastErrorMessage;
addMessage("system", lastErrorMessage);
}
}
async function stopCall(message = "Звонок остановлен.") {
if (isStopping) return;
isStopping = true;
isRunning = false;
muteMic = false;
updateMuteLabel();
if (processorNode) {
processorNode.disconnect();
processorNode.onaudioprocess = null;
}
if (sourceNode) sourceNode.disconnect();
if (mediaStream) mediaStream.getTracks().forEach((track) => track.stop());
if (socket && socket.readyState === WebSocket.OPEN) socket.close();
if (audioContext && audioContext.state !== "closed") await audioContext.close();
processorNode = null;
sourceNode = null;
mediaStream = null;
audioContext = null;
socket = null;
currentAssistantLine = null;
currentAssistantText = "";
currentAssistantAudioChunks = 0;
currentAssistantTtsFailed = false;
currentAssistantTtsMessage = "";
playbackCursor = 0;
pendingEncodedAudioCount = 0;
activeEncodedAudios.forEach((audio) => {
audio.pause();
audio.src = "";
});
activeEncodedAudios.clear();
encodedPlaybackChain = Promise.resolve();
startBtn.disabled = false;
stopBtn.disabled = true;
resetBtn.disabled = true;
partialLine.textContent = message;
levelEl.style.width = "0%";
drawIdleWave();
isStopping = false;
}
function resetSession() {
messagesEl.innerHTML = "";
currentAssistantLine = null;
currentAssistantText = "";
currentAssistantAudioChunks = 0;
currentAssistantTtsFailed = false;
currentAssistantTtsMessage = "";
partialLine.textContent = "Контекст очищен.";
if (socket && socket.readyState === WebSocket.OPEN) {
socket.send(JSON.stringify({ type: "reset" }));
}
}
function handleAudioProcess(event) {
const input = event.inputBuffer.getChannelData(0);
drawWave(input);
updateLevel(input);
if (!isRunning || muteMic || !socket || socket.readyState !== WebSocket.OPEN) return;
const downsampled = downsample(input, audioContext.sampleRate, TARGET_SAMPLE_RATE);
const pcm = floatTo16BitPcm(downsampled);
socket.send(pcm);
}
function handleServerMessage(event) {
const data = JSON.parse(event.data);
if (data.type === "ready") {
setStatus(data.model || "ready", "live");
} else if (data.type === "stt_ready") {
partialLine.textContent = "Слушаю...";
} else if (data.type === "stt_partial") {
partialLine.textContent = data.text || "Слушаю...";
} else if (data.type === "user_final") {
currentAssistantLine = null;
addMessage("user", data.text);
partialLine.textContent = "Думаю...";
} else if (data.type === "assistant_started") {
muteMic = true;
updateMuteLabel();
currentAssistantLine = addMessage("assistant", "");
currentAssistantText = "";
currentAssistantAudioChunks = 0;
currentAssistantTtsFailed = false;
currentAssistantTtsMessage = "";
setStatus("answering", "busy");
} else if (data.type === "assistant_delta") {
if (!currentAssistantLine) currentAssistantLine = addMessage("assistant", "");
currentAssistantLine.textContent += data.text;
currentAssistantText += data.text || "";
scrollTranscript();
} else if (data.type === "assistant_text_done") {
currentAssistantText = data.text || currentAssistantText;
partialLine.textContent = "Озвучиваю...";
} else if (data.type === "tts_audio") {
currentAssistantAudioChunks += 1;
if ((data.format || "").startsWith("mp3") || (data.mime_type || "").includes("mpeg")) {
playEncodedAudio(data.audio, data.mime_type || "audio/mpeg");
} else {
playPcmAudio(data.audio, data.sample_rate || TARGET_SAMPLE_RATE);
}
} else if (data.type === "tts_failed") {
currentAssistantTtsFailed = true;
currentAssistantTtsMessage = data.message || "ElevenLabs TTS не прислал аудио.";
partialLine.textContent = "ElevenLabs TTS недоступен, включаю резервную озвучку.";
addMessage("system", "ElevenLabs TTS недоступен, включаю резервную озвучку.");
} else if (data.type === "assistant_done") {
const fallbackStarted = maybeSpeakWithBrowserFallback();
if (!fallbackStarted) {
scheduleUnmuteAfterPlayback();
setStatus(currentAssistantTtsFailed ? "tts fallback" : "live", currentAssistantTtsFailed ? "busy" : "live");
partialLine.textContent = currentAssistantTtsFailed ? currentAssistantTtsMessage : "Слушаю...";
}
} else if (data.type === "reset_done") {
partialLine.textContent = "Контекст очищен.";
if (isRunning && socket && socket.readyState === WebSocket.OPEN) {
socket.send(JSON.stringify({ type: "start_greeting" }));
}
} else if (data.type === "error") {
lastErrorMessage = data.message || "Ошибка.";
partialLine.textContent = lastErrorMessage;
addMessage("system", lastErrorMessage);
setStatus("error", "error");
}
}
function addMessage(role, text) {
const line = document.createElement("div");
line.className = `line ${role}`;
line.textContent = text;
messagesEl.appendChild(line);
scrollTranscript();
return line;
}
function scrollTranscript() {
const transcript = document.querySelector(".transcript");
transcript.scrollTop = transcript.scrollHeight;
}
function setStatus(text, className) {
statusEl.textContent = text;
statusEl.className = `status-pill ${className || ""}`.trim();
}
function maybeSpeakWithBrowserFallback() {
if (currentAssistantAudioChunks > 0) return false;
if (!currentAssistantText.trim()) return false;
if (!("speechSynthesis" in window) || typeof SpeechSynthesisUtterance === "undefined") {
partialLine.textContent = currentAssistantTtsFailed
? currentAssistantTtsMessage
: "TTS не прислал аудио.";
return false;
}
window.speechSynthesis.cancel();
const utterance = new SpeechSynthesisUtterance(currentAssistantText);
utterance.lang = "ru-RU";
utterance.rate = 1.04;
utterance.pitch = 1;
utterance.onend = finishBrowserFallbackSpeech;
utterance.onerror = finishBrowserFallbackSpeech;
muteMic = true;
updateMuteLabel();
setStatus("browser voice", "busy");
partialLine.textContent = currentAssistantTtsFailed
? "Резервная озвучка включена: ElevenLabs TTS не дал аудио."
: "Резервная озвучка включена.";
window.speechSynthesis.speak(utterance);
return true;
}
function finishBrowserFallbackSpeech() {
muteMic = false;
updateMuteLabel();
setStatus("live", "live");
partialLine.textContent = "Слушаю...";
}
function updateMuteLabel() {
muteLabel.textContent = muteMic ? "muted" : "open";
}
function waitForSocketOpen(ws) {
return new Promise((resolve, reject) => {
ws.addEventListener("open", resolve, { once: true });
ws.addEventListener("error", reject, { once: true });
});
}
function waitForServerReady(ws) {
return new Promise((resolve, reject) => {
const timeout = window.setTimeout(() => {
cleanup();
reject(new Error("Сервер не ответил."));
}, 6000);
function cleanup() {
window.clearTimeout(timeout);
ws.removeEventListener("message", onMessage);
ws.removeEventListener("close", onClose);
ws.removeEventListener("error", onError);
}
function onMessage(event) {
const data = JSON.parse(event.data);
if (data.type === "ready") {
cleanup();
resolve(data);
} else if (data.type === "error") {
cleanup();
reject(new Error(data.message || "Сервер вернул ошибку."));
}
}
function onClose() {
cleanup();
reject(new Error("WebSocket закрыт."));
}
function onError() {
cleanup();
reject(new Error("Ошибка WebSocket."));
}
ws.addEventListener("message", onMessage);
ws.addEventListener("close", onClose);
ws.addEventListener("error", onError);
});
}
async function cleanupAfterFailedStart() {
if (processorNode) {
processorNode.disconnect();
processorNode.onaudioprocess = null;
}
if (sourceNode) sourceNode.disconnect();
if (mediaStream) mediaStream.getTracks().forEach((track) => track.stop());
if (socket && socket.readyState === WebSocket.OPEN) socket.close();
if (audioContext) await audioContext.close();
processorNode = null;
sourceNode = null;
mediaStream = null;
audioContext = null;
socket = null;
}
function downsample(buffer, inputRate, outputRate) {
if (outputRate === inputRate) return buffer;
const ratio = inputRate / outputRate;
const newLength = Math.round(buffer.length / ratio);
const result = new Float32Array(newLength);
let offsetResult = 0;
let offsetBuffer = 0;
while (offsetResult < result.length) {
const nextOffsetBuffer = Math.round((offsetResult + 1) * ratio);
let accumulator = 0;
let count = 0;
for (let i = offsetBuffer; i < nextOffsetBuffer && i < buffer.length; i += 1) {
accumulator += buffer[i];
count += 1;
}
result[offsetResult] = accumulator / Math.max(count, 1);
offsetResult += 1;
offsetBuffer = nextOffsetBuffer;
}
return result;
}
function floatTo16BitPcm(floatBuffer) {
const output = new Int16Array(floatBuffer.length);
for (let i = 0; i < floatBuffer.length; i += 1) {
const sample = Math.max(-1, Math.min(1, floatBuffer[i]));
output[i] = sample < 0 ? sample * 0x8000 : sample * 0x7fff;
}
return output.buffer;
}
function playPcmAudio(base64Audio, sampleRate) {
if (!audioContext) return;
if (audioContext.state === "suspended") audioContext.resume();
const bytes = Uint8Array.from(atob(base64Audio), (char) => char.charCodeAt(0));
const samples = new Int16Array(bytes.buffer);
const audioBuffer = audioContext.createBuffer(1, samples.length, sampleRate);
const channel = audioBuffer.getChannelData(0);
for (let i = 0; i < samples.length; i += 1) {
channel[i] = samples[i] / 32768;
}
const source = audioContext.createBufferSource();
source.buffer = audioBuffer;
source.connect(audioContext.destination);
const startAt = Math.max(audioContext.currentTime + 0.03, playbackCursor);
source.start(startAt);
playbackCursor = startAt + audioBuffer.duration;
}
function playEncodedAudio(base64Audio, mimeType) {
pendingEncodedAudioCount += 1;
encodedPlaybackChain = encodedPlaybackChain.then(
() =>
new Promise((resolve) => {
const bytes = Uint8Array.from(atob(base64Audio), (char) => char.charCodeAt(0));
const blob = new Blob([bytes], { type: mimeType });
const url = URL.createObjectURL(blob);
const audio = new Audio(url);
activeEncodedAudios.add(audio);
function cleanup() {
URL.revokeObjectURL(url);
activeEncodedAudios.delete(audio);
pendingEncodedAudioCount = Math.max(0, pendingEncodedAudioCount - 1);
if (pendingEncodedAudioCount === 0 && muteMic) {
scheduleUnmuteAfterPlayback();
}
resolve();
}
audio.onended = cleanup;
audio.onerror = cleanup;
audio.play().catch(cleanup);
}),
);
}
function scheduleUnmuteAfterPlayback() {
if (pendingEncodedAudioCount > 0) return;
if (pendingUnmuteTimer) window.clearTimeout(pendingUnmuteTimer);
const remainingMs = audioContext ? Math.max(0, (playbackCursor - audioContext.currentTime) * 1000) : 0;
pendingUnmuteTimer = window.setTimeout(() => {
muteMic = false;
updateMuteLabel();
}, remainingMs + 120);
}
function updateLevel(buffer) {
let sum = 0;
for (let i = 0; i < buffer.length; i += 1) sum += buffer[i] * buffer[i];
const rms = Math.sqrt(sum / buffer.length);
const percent = Math.min(100, Math.round(rms * 260));
levelEl.style.width = `${percent}%`;
}
function drawIdleWave() {
canvasCtx.fillStyle = "#101820";
canvasCtx.fillRect(0, 0, canvas.width, canvas.height);
canvasCtx.strokeStyle = "#2f9e80";
canvasCtx.lineWidth = 3;
canvasCtx.beginPath();
const mid = canvas.height / 2;
for (let x = 0; x < canvas.width; x += 1) {
const y = mid + Math.sin(x / 28) * 10 + Math.sin(x / 83) * 18;
if (x === 0) canvasCtx.moveTo(x, y);
else canvasCtx.lineTo(x, y);
}
canvasCtx.stroke();
}
function drawWave(buffer) {
canvasCtx.fillStyle = "#101820";
canvasCtx.fillRect(0, 0, canvas.width, canvas.height);
canvasCtx.strokeStyle = muteMic ? "#d9563f" : "#36c28f";
canvasCtx.lineWidth = 3;
canvasCtx.beginPath();
const slice = Math.max(1, Math.floor(buffer.length / canvas.width));
const mid = canvas.height / 2;
for (let x = 0; x < canvas.width; x += 1) {
const sample = buffer[x * slice] || 0;
const y = mid + sample * mid * 0.84;
if (x === 0) canvasCtx.moveTo(x, y);
else canvasCtx.lineTo(x, y);
}
canvasCtx.stroke();
}