mirror of
https://github.com/TaterTotterson/microWakeWord-Trainer-Nvidia-Docker.git
synced 2026-06-13 04:20:19 -06:00
d5dcfbf5f1
- Add LANGUAGE default (en) to shell.functions - setup_python_venv downloads Dutch ONNX voices (pim, ronnie, nathalie) - wake_word_sample_generator uses multiple --model flags for single-speaker voices, cycling between them for variety - train_wake_word accepts and passes --language through the pipeline - recorder_server.py accepts language in session API - Web UI adds language dropdown (English/Dutch)
818 lines
25 KiB
HTML
818 lines
25 KiB
HTML
<!doctype html>
|
||
<html>
|
||
<head>
|
||
<meta charset="utf-8" />
|
||
<meta name="viewport" content="width=device-width, initial-scale=1" />
|
||
<title>microWakeWord Recorder</title>
|
||
<style>
|
||
:root{
|
||
--bg: #070709;
|
||
--panel: rgba(18, 18, 22, 0.78);
|
||
--panel2: rgba(24, 24, 30, 0.86);
|
||
--text: #e9e9ee;
|
||
--muted: #a2a2ad;
|
||
--line: rgba(255,255,255,0.10);
|
||
--orange: #ff8a2a;
|
||
--orange2:#ffb066;
|
||
--ok:#38d39f;
|
||
--warn:#ffb020;
|
||
--err:#ff4a4a;
|
||
--shadow: 0 18px 50px rgba(0,0,0,0.45);
|
||
--radius: 16px;
|
||
}
|
||
|
||
html, body { height: 100%; }
|
||
body {
|
||
margin: 0;
|
||
color: var(--text);
|
||
font-family: ui-sans-serif, system-ui, -apple-system, Segoe UI, Roboto, sans-serif;
|
||
background:
|
||
radial-gradient(900px 500px at 12% 6%, rgba(255, 138, 42, 0.12), transparent 55%),
|
||
radial-gradient(700px 420px at 80% 14%, rgba(255, 176, 102, 0.09), transparent 60%),
|
||
radial-gradient(800px 600px at 50% 100%, rgba(255, 138, 42, 0.06), transparent 55%),
|
||
linear-gradient(180deg, #050506 0%, #09090d 100%);
|
||
}
|
||
|
||
.wrap { max-width: 940px; margin: 0 auto; padding: 26px 18px 42px; }
|
||
|
||
h2 { margin: 0 0 8px; font-size: 22px; letter-spacing: 0.2px; }
|
||
p { margin: 0 0 14px; color: var(--muted); line-height: 1.45; }
|
||
|
||
.topbar {
|
||
display:flex; align-items:center; justify-content:space-between;
|
||
gap: 12px; margin-bottom: 14px;
|
||
}
|
||
|
||
.brand { display:flex; align-items:center; gap:10px; }
|
||
.logo {
|
||
width: 38px; height: 38px; border-radius: 12px;
|
||
background:
|
||
radial-gradient(circle at 30% 30%, rgba(255,176,102,0.55), rgba(255,138,42,0.25) 45%, rgba(0,0,0,0) 72%),
|
||
linear-gradient(180deg, rgba(255,138,42,0.22), rgba(255,138,42,0.06));
|
||
border: 1px solid rgba(255,138,42,0.30);
|
||
box-shadow: 0 10px 28px rgba(255,138,42,0.08);
|
||
}
|
||
|
||
.row { display: flex; gap: 12px; flex-wrap: wrap; align-items: center; }
|
||
|
||
.card {
|
||
border: 1px solid var(--line);
|
||
background: linear-gradient(180deg, var(--panel), var(--panel2));
|
||
border-radius: var(--radius);
|
||
padding: 16px;
|
||
margin-top: 14px;
|
||
box-shadow: var(--shadow);
|
||
backdrop-filter: blur(8px);
|
||
}
|
||
|
||
.muted { color: var(--muted); }
|
||
|
||
input[type="text"], input[type="number"]{
|
||
padding: 11px 12px;
|
||
font-size: 15px;
|
||
border-radius: 12px;
|
||
border: 1px solid rgba(255,255,255,0.12);
|
||
background: rgba(0,0,0,0.35);
|
||
color: var(--text);
|
||
outline: none;
|
||
}
|
||
input[type="text"] { width: 420px; max-width: 100%; }
|
||
input[type="number"] { width: 120px; }
|
||
input::placeholder { color: rgba(233,233,238,0.35); }
|
||
|
||
button {
|
||
padding: 10px 14px;
|
||
font-size: 13px;
|
||
cursor: pointer;
|
||
border-radius: 12px;
|
||
border: 1px solid rgba(255,255,255,0.14);
|
||
background: rgba(255,255,255,0.06);
|
||
color: var(--text);
|
||
transition: transform 0.04s ease, border-color .15s ease, background .15s ease;
|
||
}
|
||
button:hover { border-color: rgba(255,138,42,0.35); background: rgba(255,255,255,0.08); }
|
||
button:active { transform: translateY(1px); }
|
||
button:disabled { opacity: 0.45; cursor: not-allowed; }
|
||
|
||
.primary {
|
||
border-color: rgba(255,138,42,0.40);
|
||
background: linear-gradient(180deg, rgba(255,138,42,0.24), rgba(255,138,42,0.12));
|
||
}
|
||
.primary:hover { border-color: rgba(255,138,42,0.65); }
|
||
|
||
.pill {
|
||
display:inline-block;
|
||
padding: 4px 10px;
|
||
border-radius: 999px;
|
||
background: rgba(255,255,255,0.07);
|
||
border: 1px solid rgba(255,255,255,0.10);
|
||
color: var(--muted);
|
||
font-size: 12px;
|
||
}
|
||
.pill.ok { color: var(--ok); border-color: rgba(56,211,159,0.25); background: rgba(56,211,159,0.08); }
|
||
.pill.warn { color: var(--warn); border-color: rgba(255,176,32,0.25); background: rgba(255,176,32,0.08); }
|
||
.pill.err { color: var(--err); border-color: rgba(255,74,74,0.25); background: rgba(255,74,74,0.08); }
|
||
|
||
details { margin-top: 10px; }
|
||
summary { cursor: pointer; color: var(--orange2); }
|
||
summary:hover { color: var(--orange); }
|
||
|
||
label { display:flex; gap:10px; align-items:center; }
|
||
input[type="range"] { width: 240px; }
|
||
|
||
.meter {
|
||
height: 10px;
|
||
background: rgba(255,255,255,0.08);
|
||
border-radius: 999px;
|
||
overflow: hidden;
|
||
width: 280px;
|
||
border: 1px solid rgba(255,255,255,0.10);
|
||
}
|
||
.meter > div {
|
||
height: 10px;
|
||
width: 0%;
|
||
background: linear-gradient(90deg, rgba(255,138,42,0.55), rgba(255,176,102,0.85));
|
||
}
|
||
|
||
pre {
|
||
background: rgba(0,0,0,0.55);
|
||
color: #e6e6ea;
|
||
padding: 12px;
|
||
border-radius: 14px;
|
||
overflow: auto;
|
||
max-height: 300px;
|
||
border: 1px solid rgba(255,255,255,0.10);
|
||
white-space: pre-wrap;
|
||
word-break: break-word;
|
||
}
|
||
|
||
.big { font-size: 16px; }
|
||
|
||
.divider {
|
||
height: 1px;
|
||
width: 100%;
|
||
background: rgba(255,255,255,0.10);
|
||
margin: 12px 0;
|
||
}
|
||
</style>
|
||
</head>
|
||
|
||
<body>
|
||
<div class="wrap">
|
||
<div class="topbar">
|
||
<div class="brand">
|
||
<div class="logo"></div>
|
||
<div>
|
||
<h2>🎙️ microWakeWord Personal Recorder</h2>
|
||
<p class="muted">Enter a wake word, test TTS pronunciation, then record takes. Recording starts when you speak and stops after silence.</p>
|
||
</div>
|
||
</div>
|
||
</div>
|
||
|
||
<div class="card">
|
||
<div class="row">
|
||
<input id="phrase" type="text" placeholder='e.g. "tater totterson"' />
|
||
<button id="startSessionBtn" class="primary">Start session</button>
|
||
<button id="ttsBtn" disabled>🔊 Test TTS</button>
|
||
<span id="sessionPill" class="pill">No session</span>
|
||
</div>
|
||
|
||
<div class="row" style="margin-top:10px;">
|
||
<label class="muted">Language
|
||
<select id="language" style="padding:10px 12px;font-size:15px;border-radius:12px;border:1px solid rgba(255,255,255,0.12);background:rgba(0,0,0,0.35);color:var(--text);outline:none;">
|
||
<option value="en" selected>English (en)</option>
|
||
<option value="nl">Dutch (nl)</option>
|
||
</select>
|
||
</label>
|
||
<label class="muted">Speakers
|
||
<input id="speakersTotal" type="number" min="1" max="10" value="1" />
|
||
</label>
|
||
<label class="muted">Takes / speaker
|
||
<input id="takesPerSpeaker" type="number" min="1" max="50" value="10" />
|
||
</label>
|
||
<span id="speakerPill" class="pill">Speaker: -</span>
|
||
</div>
|
||
|
||
<details>
|
||
<summary>Advanced (if it’s too sensitive / not sensitive enough)</summary>
|
||
<div style="margin-top:10px;">
|
||
<label>
|
||
Start sensitivity
|
||
<input id="startThresh" type="range" min="0.005" max="0.08" step="0.001" value="0.02" />
|
||
<span id="startThreshVal" class="muted"></span>
|
||
</label>
|
||
<label>
|
||
Silence stop (ms)
|
||
<input id="silenceMs" type="range" min="300" max="2000" step="50" value="900" />
|
||
<span id="silenceMsVal" class="muted"></span>
|
||
</label>
|
||
<label>
|
||
Min take length (ms)
|
||
<input id="minTakeMs" type="range" min="300" max="2000" step="50" value="650" />
|
||
<span id="minTakeMsVal" class="muted"></span>
|
||
</label>
|
||
</div>
|
||
</details>
|
||
</div>
|
||
|
||
<div class="card">
|
||
<div class="row">
|
||
<button id="beginBtn" disabled class="primary">🎬 Begin recording</button>
|
||
<button id="resetBtn" disabled>🧹 Reset recordings</button>
|
||
<button id="trainBtn" disabled>🧠 Start training</button>
|
||
<span id="status" class="pill">Idle</span>
|
||
</div>
|
||
|
||
<div style="margin-top:12px;" class="row">
|
||
<div class="meter"><div id="meterFill"></div></div>
|
||
<span class="muted" id="meterText">Mic level</span>
|
||
</div>
|
||
|
||
<div class="divider"></div>
|
||
|
||
<p class="big">
|
||
Speaker: <b id="speakerNum">-</b> / <b id="speakerTotal">-</b>
|
||
<span id="speakerState" class="pill">Waiting</span>
|
||
</p>
|
||
|
||
<p class="big">
|
||
Take: <b id="takeNum">0</b> / <b id="takeTotal">10</b>
|
||
<span id="takeState" class="pill">Not recording</span>
|
||
</p>
|
||
|
||
<div id="takesList" class="muted"></div>
|
||
|
||
<h4 style="margin-top: 18px; margin-bottom: 10px;">Training log</h4>
|
||
<pre id="trainLog">(no training started)</pre>
|
||
</div>
|
||
</div>
|
||
|
||
<script>
|
||
const $ = (id) => document.getElementById(id);
|
||
|
||
function setPill(el, text, cls) {
|
||
el.className = "pill " + (cls || "");
|
||
el.textContent = text;
|
||
}
|
||
|
||
async function api(path, opts) {
|
||
opts = opts || {};
|
||
// Always try to avoid cache for polling endpoints
|
||
if (!opts.cache) opts.cache = "no-store";
|
||
|
||
const res = await fetch(path, opts);
|
||
const ct = res.headers.get("content-type") || "";
|
||
const data = ct.includes("application/json") ? await res.json() : await res.text();
|
||
if (!res.ok) {
|
||
const err = (typeof data === "string") ? { error: data } : (data || {});
|
||
const msg = err.error || err.message || JSON.stringify(err);
|
||
const e = new Error(msg);
|
||
e.details = err;
|
||
throw e;
|
||
}
|
||
return data;
|
||
}
|
||
|
||
// -------------------- log auto-scroll (sticky to bottom) --------------------
|
||
function isNearBottom(el, px = 40) {
|
||
return (el.scrollHeight - el.scrollTop - el.clientHeight) <= px;
|
||
}
|
||
|
||
function setLogTextAutoScroll(el, text) {
|
||
const stick = isNearBottom(el);
|
||
el.textContent = text || "";
|
||
if (stick) el.scrollTop = el.scrollHeight;
|
||
}
|
||
// --------------------------------------------------------------------------
|
||
|
||
let session = null;
|
||
let isRunning = false;
|
||
|
||
let stream = null;
|
||
let audioCtx = null;
|
||
let analyser = null;
|
||
let source = null;
|
||
|
||
let capturing = false;
|
||
let startedAt = 0;
|
||
let silenceStart = null;
|
||
let floatChunks = [];
|
||
let frameSize = 2048;
|
||
|
||
let currentSpeaker = 1;
|
||
let speakersTotal = 1;
|
||
|
||
let currentTake = 0;
|
||
let takesPerSpeaker = 10;
|
||
|
||
// --- training poll (append mode; scrollback works) ---
|
||
let trainingPollRunning = false;
|
||
let trainingPollAbort = false;
|
||
|
||
let logBuffer = ""; // full text we’ve shown in the browser
|
||
let lastChunk = ""; // last chunk we received (for de-dupe)
|
||
let seenAnyOutput = false;
|
||
|
||
function appendLogAutoScroll(el, chunk) {
|
||
if (!chunk) return;
|
||
const stick = isNearBottom(el);
|
||
el.textContent += chunk;
|
||
if (stick) el.scrollTop = el.scrollHeight;
|
||
}
|
||
|
||
function startThreshold() { return parseFloat($("startThresh").value); }
|
||
function silenceStopMs() { return parseInt($("silenceMs").value, 10); }
|
||
function minTakeMs() { return parseInt($("minTakeMs").value, 10); }
|
||
|
||
function updateAdvancedLabels() {
|
||
$("startThreshVal").textContent = startThreshold().toFixed(3);
|
||
$("silenceMsVal").textContent = silenceStopMs() + "ms";
|
||
$("minTakeMsVal").textContent = minTakeMs() + "ms";
|
||
}
|
||
["startThresh","silenceMs","minTakeMs"].forEach(id => $(id).addEventListener("input", updateAdvancedLabels));
|
||
updateAdvancedLabels();
|
||
|
||
function refreshUI() {
|
||
$("speakerNum").textContent = String(currentSpeaker);
|
||
$("speakerTotal").textContent = String(speakersTotal);
|
||
$("takeNum").textContent = String(currentTake);
|
||
$("takeTotal").textContent = String(takesPerSpeaker);
|
||
setPill($("speakerPill"), `Speaker ${currentSpeaker}/${speakersTotal}`);
|
||
}
|
||
|
||
// -------------------- mic lifecycle --------------------
|
||
async function ensureMic() {
|
||
if (!navigator.mediaDevices || !navigator.mediaDevices.getUserMedia) {
|
||
throw new Error("Microphone not available here. Use https:// (or http://localhost) to record.");
|
||
}
|
||
if (stream) return;
|
||
stream = await navigator.mediaDevices.getUserMedia({ audio: true, video: false });
|
||
audioCtx = new (window.AudioContext || window.webkitAudioContext)();
|
||
analyser = audioCtx.createAnalyser();
|
||
analyser.fftSize = 2048;
|
||
source = audioCtx.createMediaStreamSource(stream);
|
||
source.connect(analyser);
|
||
requestAnimationFrame(meterLoop);
|
||
}
|
||
|
||
async function stopMicNow() {
|
||
isRunning = false;
|
||
capturing = false;
|
||
|
||
const proc = window.__mw_proc;
|
||
if (proc) {
|
||
try { proc.disconnect(); } catch {}
|
||
try { source && source.disconnect(proc); } catch {}
|
||
window.__mw_proc = null;
|
||
}
|
||
|
||
if (stream) {
|
||
try { stream.getTracks().forEach(t => t.stop()); } catch {}
|
||
stream = null;
|
||
}
|
||
|
||
if (audioCtx) {
|
||
try { await audioCtx.close(); } catch {}
|
||
audioCtx = null;
|
||
}
|
||
|
||
analyser = null;
|
||
source = null;
|
||
|
||
$("meterFill").style.width = "0%";
|
||
$("meterText").textContent = "Mic stopped";
|
||
}
|
||
|
||
function meterLoop() {
|
||
if (!analyser) {
|
||
requestAnimationFrame(meterLoop);
|
||
return;
|
||
}
|
||
|
||
const data = new Uint8Array(analyser.fftSize);
|
||
analyser.getByteTimeDomainData(data);
|
||
|
||
let sumSq = 0;
|
||
for (let i=0;i<data.length;i++){
|
||
const v = (data[i] - 128) / 128;
|
||
sumSq += v*v;
|
||
}
|
||
const rms = Math.sqrt(sumSq / data.length);
|
||
const pct = Math.min(100, Math.max(0, rms * 600));
|
||
$("meterFill").style.width = pct + "%";
|
||
$("meterText").textContent = `Mic level (rms=${rms.toFixed(3)})`;
|
||
|
||
if (isRunning) recorderTick(rms);
|
||
|
||
requestAnimationFrame(meterLoop);
|
||
}
|
||
|
||
// -------------------- recording state machine --------------------
|
||
function recorderTick(rms) {
|
||
const now = performance.now();
|
||
|
||
if (!capturing) {
|
||
if (rms >= startThreshold()) startCapture();
|
||
return;
|
||
}
|
||
|
||
if (rms < startThreshold() * 0.65) {
|
||
if (silenceStart === null) silenceStart = now;
|
||
const silentFor = now - silenceStart;
|
||
if (silentFor >= silenceStopMs()) {
|
||
const dur = now - startedAt;
|
||
if (dur >= minTakeMs()) stopCaptureAndUpload();
|
||
else silenceStart = now;
|
||
}
|
||
} else {
|
||
silenceStart = null;
|
||
}
|
||
}
|
||
|
||
async function startCapture() {
|
||
capturing = true;
|
||
startedAt = performance.now();
|
||
silenceStart = null;
|
||
floatChunks = [];
|
||
|
||
setPill($("takeState"), "Recording…", "warn");
|
||
|
||
const proc = audioCtx.createScriptProcessor(frameSize, 1, 1);
|
||
source.connect(proc);
|
||
proc.connect(audioCtx.destination);
|
||
|
||
proc.onaudioprocess = (ev) => {
|
||
if (!capturing) return;
|
||
const chan = ev.inputBuffer.getChannelData(0);
|
||
floatChunks.push(new Float32Array(chan));
|
||
};
|
||
|
||
window.__mw_proc = proc;
|
||
}
|
||
|
||
async function stopCaptureAndUpload() {
|
||
capturing = false;
|
||
setPill($("takeState"), "Processing…");
|
||
|
||
const proc = window.__mw_proc;
|
||
if (proc) {
|
||
try { proc.disconnect(); } catch {}
|
||
try { source.disconnect(proc); } catch {}
|
||
window.__mw_proc = null;
|
||
}
|
||
|
||
currentTake += 1;
|
||
refreshUI();
|
||
|
||
let totalLen = 0;
|
||
for (const c of floatChunks) totalLen += c.length;
|
||
const merged = new Float32Array(totalLen);
|
||
let off = 0;
|
||
for (const c of floatChunks) { merged.set(c, off); off += c.length; }
|
||
|
||
const wavBlob = await floatToWav16kMono(merged, audioCtx.sampleRate);
|
||
|
||
try {
|
||
setPill($("status"), `Uploading speaker ${currentSpeaker} take ${currentTake}…`, "warn");
|
||
|
||
const fd = new FormData();
|
||
fd.append("speaker_index", String(currentSpeaker));
|
||
fd.append("take_index", String(currentTake));
|
||
fd.append("file", wavBlob, `take_${String(currentTake).padStart(2,"0")}.wav`);
|
||
|
||
await api("/api/upload_take", { method:"POST", body: fd });
|
||
|
||
$("takesList").textContent = `Saved ${currentTake}/${takesPerSpeaker} takes for speaker ${currentSpeaker}/${speakersTotal}`;
|
||
setPill($("status"), `Saved speaker ${currentSpeaker} take ${currentTake}/${takesPerSpeaker}`, "ok");
|
||
|
||
if (currentTake >= takesPerSpeaker) {
|
||
if (currentSpeaker >= speakersTotal) {
|
||
setPill($("takeState"), "Done", "ok");
|
||
setPill($("speakerState"), "All speakers done ✅", "ok");
|
||
setPill($("status"), "All takes recorded ✅", "ok");
|
||
|
||
await stopMicNow();
|
||
await autoStartTraining();
|
||
return;
|
||
}
|
||
|
||
currentSpeaker += 1;
|
||
currentTake = 0;
|
||
refreshUI();
|
||
|
||
setPill($("speakerState"), `Speaker ${currentSpeaker - 1} complete ✅`, "ok");
|
||
setPill($("takeState"), "Paused", "warn");
|
||
setPill($("status"), `Ready for speaker ${currentSpeaker}. Click Begin recording.`, "warn");
|
||
|
||
isRunning = false;
|
||
$("beginBtn").disabled = false;
|
||
|
||
await stopMicNow();
|
||
return;
|
||
}
|
||
|
||
setPill($("speakerState"), `Speaker ${currentSpeaker}/${speakersTotal}`);
|
||
setPill($("takeState"), "Listening…", "ok");
|
||
|
||
} catch (e) {
|
||
console.error(e);
|
||
setPill($("status"), "Upload failed", "err");
|
||
setPill($("takeState"), "Error", "err");
|
||
isRunning = false;
|
||
$("beginBtn").disabled = false;
|
||
alert("Upload failed: " + e.message);
|
||
}
|
||
}
|
||
|
||
// -------------------- WAV encoding helpers --------------------
|
||
async function floatToWav16kMono(float32, srcRate) {
|
||
const buf = audioCtx.createBuffer(1, float32.length, srcRate);
|
||
buf.copyToChannel(float32, 0);
|
||
|
||
const targetRate = 16000;
|
||
const targetLen = Math.max(1, Math.round(float32.length * targetRate / srcRate));
|
||
const offline = new OfflineAudioContext(1, targetLen, targetRate);
|
||
|
||
const src = offline.createBufferSource();
|
||
src.buffer = buf;
|
||
src.connect(offline.destination);
|
||
src.start(0);
|
||
|
||
const rendered = await offline.startRendering();
|
||
const data = rendered.getChannelData(0);
|
||
|
||
const wav = encodeWavPCM16(data, targetRate);
|
||
return new Blob([wav], { type: "audio/wav" });
|
||
}
|
||
|
||
function encodeWavPCM16(float32, sampleRate) {
|
||
const numSamples = float32.length;
|
||
const buffer = new ArrayBuffer(44 + numSamples * 2);
|
||
const view = new DataView(buffer);
|
||
|
||
function writeString(offset, str) {
|
||
for (let i=0;i<str.length;i++) view.setUint8(offset+i, str.charCodeAt(i));
|
||
}
|
||
|
||
writeString(0, "RIFF");
|
||
view.setUint32(4, 36 + numSamples * 2, true);
|
||
writeString(8, "WAVE");
|
||
|
||
writeString(12, "fmt ");
|
||
view.setUint32(16, 16, true);
|
||
view.setUint16(20, 1, true);
|
||
view.setUint16(22, 1, true);
|
||
view.setUint32(24, sampleRate, true);
|
||
view.setUint32(28, sampleRate * 2, true);
|
||
view.setUint16(32, 2, true);
|
||
view.setUint16(34, 16, true);
|
||
|
||
writeString(36, "data");
|
||
view.setUint32(40, numSamples * 2, true);
|
||
|
||
let offset = 44;
|
||
for (let i=0;i<numSamples;i++) {
|
||
let s = Math.max(-1, Math.min(1, float32[i]));
|
||
const v = s < 0 ? s * 0x8000 : s * 0x7fff;
|
||
view.setInt16(offset, v, true);
|
||
offset += 2;
|
||
}
|
||
return buffer;
|
||
}
|
||
|
||
// -------------------- training (manual + auto) --------------------
|
||
async function startTrainingWithPrompt(auto=false) {
|
||
const sess = await api("/api/session", { method: "GET" });
|
||
const takesReceived = sess.takes_received || 0;
|
||
const total = (sess.speakers_total || 1) * (sess.takes_per_speaker || 10);
|
||
|
||
let allowNoPersonal = false;
|
||
|
||
if (takesReceived === 0) {
|
||
const ok = confirm(
|
||
`No personal voice samples recorded (0/${total}).\n\nTrain anyway WITHOUT personal voices?`
|
||
);
|
||
if (!ok) return;
|
||
allowNoPersonal = true;
|
||
}
|
||
|
||
// lock UI immediately
|
||
$("trainBtn").disabled = true;
|
||
$("beginBtn").disabled = true;
|
||
$("resetBtn").disabled = true;
|
||
|
||
setPill($("status"), auto ? "Auto-starting training…" : "Preparing training environment…", "warn");
|
||
|
||
// Reset log state for a fresh run
|
||
trainingPollAbort = false;
|
||
logBuffer = "";
|
||
lastChunk = "";
|
||
seenAnyOutput = false;
|
||
|
||
const logEl = $("trainLog");
|
||
logEl.textContent = "(preparing…)\n";
|
||
|
||
try {
|
||
// Kick off training first
|
||
await api("/api/train", {
|
||
method: "POST",
|
||
headers: { "Content-Type": "application/json" },
|
||
body: JSON.stringify({ allow_no_personal: allowNoPersonal })
|
||
});
|
||
|
||
// Only start polling AFTER training was successfully kicked off
|
||
if (!trainingPollRunning) {
|
||
trainingPollRunning = true;
|
||
pollTrainingTail();
|
||
}
|
||
|
||
setPill($("status"), "Training running…", "warn");
|
||
} catch (e) {
|
||
$("trainBtn").disabled = false;
|
||
$("resetBtn").disabled = false;
|
||
$("beginBtn").disabled = false;
|
||
trainingPollAbort = true;
|
||
trainingPollRunning = false;
|
||
throw e;
|
||
}
|
||
}
|
||
|
||
async function autoStartTraining() {
|
||
try {
|
||
await startTrainingWithPrompt(true);
|
||
} catch (e) {
|
||
console.error(e);
|
||
setPill($("status"), "Auto-train failed", "err");
|
||
alert("Auto-start training failed: " + e.message);
|
||
}
|
||
}
|
||
|
||
$("trainBtn").addEventListener("click", async () => {
|
||
try {
|
||
await startTrainingWithPrompt(false);
|
||
} catch (e) {
|
||
alert("Train failed: " + e.message);
|
||
setPill($("status"), "Train failed", "err");
|
||
}
|
||
});
|
||
|
||
async function pollTrainingTail() {
|
||
const logEl = $("trainLog");
|
||
|
||
for (;;) {
|
||
if (trainingPollAbort) {
|
||
trainingPollRunning = false;
|
||
break;
|
||
}
|
||
|
||
try {
|
||
const st = await api(`/api/train_status?ts=${Date.now()}`, { method:"GET", cache:"no-store" });
|
||
const tr = st.training || {};
|
||
|
||
// NOTE: this assumes /api/train_status returns NEW output chunks (not full tail snapshots)
|
||
const chunkRaw = tr.log_text || "";
|
||
const chunk = chunkRaw; // keep exact newlines from server
|
||
|
||
if (chunk) {
|
||
// wipe placeholder once
|
||
if (!seenAnyOutput) {
|
||
logEl.textContent = "";
|
||
logBuffer = "";
|
||
lastChunk = "";
|
||
seenAnyOutput = true;
|
||
}
|
||
|
||
// simple de-dupe: if server repeats the same chunk, skip it
|
||
if (chunk !== lastChunk) {
|
||
lastChunk = chunk;
|
||
logBuffer += chunk;
|
||
appendLogAutoScroll(logEl, chunk);
|
||
}
|
||
} else {
|
||
// before first output, show waiting message but do NOT overwrite later scrollback
|
||
if (!seenAnyOutput) {
|
||
if (!logEl.textContent || logEl.textContent.includes("(no training") || logEl.textContent.startsWith("(preparing…")) {
|
||
logEl.textContent = "Waiting for training output…\n";
|
||
}
|
||
}
|
||
}
|
||
|
||
const exitCodeIsSet = (tr.exit_code !== null && tr.exit_code !== undefined);
|
||
|
||
if (!tr.running && exitCodeIsSet) {
|
||
$("trainBtn").disabled = false;
|
||
$("resetBtn").disabled = false;
|
||
$("beginBtn").disabled = false;
|
||
|
||
if (tr.exit_code === 0) setPill($("status"), "Training finished ✅", "ok");
|
||
else setPill($("status"), `Training ended (exit=${tr.exit_code})`, "err");
|
||
|
||
trainingPollRunning = false;
|
||
break;
|
||
}
|
||
} catch (e) {
|
||
// ignore transient polling errors
|
||
}
|
||
|
||
await new Promise(r => setTimeout(r, 1000));
|
||
}
|
||
}
|
||
|
||
// -------------------- session + UI wiring --------------------
|
||
$("ttsBtn").addEventListener("click", () => {
|
||
const phrase = ($("phrase").value || "").trim();
|
||
if (!phrase) return;
|
||
const u = new SpeechSynthesisUtterance(phrase);
|
||
speechSynthesis.cancel();
|
||
speechSynthesis.speak(u);
|
||
});
|
||
|
||
$("startSessionBtn").addEventListener("click", async () => {
|
||
const phrase = ($("phrase").value || "").trim();
|
||
if (!phrase) { alert("Enter a wake word phrase first."); return; }
|
||
|
||
speakersTotal = parseInt($("speakersTotal").value || "1", 10);
|
||
takesPerSpeaker = parseInt($("takesPerSpeaker").value || "10", 10);
|
||
const language = $("language").value || "en";
|
||
|
||
try {
|
||
setPill($("sessionPill"), "Starting…", "warn");
|
||
const data = await api("/api/start_session", {
|
||
method: "POST",
|
||
headers: {"Content-Type":"application/json"},
|
||
body: JSON.stringify({ phrase, speakers_total: speakersTotal, takes_per_speaker: takesPerSpeaker, language })
|
||
});
|
||
|
||
session = data;
|
||
|
||
currentSpeaker = 1;
|
||
currentTake = 0;
|
||
|
||
$("takesList").textContent = "";
|
||
$("trainLog").textContent = "(no training started)";
|
||
|
||
// Stop any previous poll loop cleanly
|
||
trainingPollAbort = true;
|
||
trainingPollRunning = false;
|
||
logBuffer = "";
|
||
lastChunk = "";
|
||
seenAnyOutput = false;
|
||
|
||
refreshUI();
|
||
|
||
await stopMicNow();
|
||
|
||
setPill($("sessionPill"), `Session: ${data.safe_word} (${data.language || "en"})`, "ok");
|
||
$("beginBtn").disabled = false;
|
||
$("resetBtn").disabled = false;
|
||
$("trainBtn").disabled = false;
|
||
$("ttsBtn").disabled = false;
|
||
|
||
setPill($("status"), "Ready", "ok");
|
||
setPill($("speakerState"), "Waiting");
|
||
setPill($("takeState"), "Not recording");
|
||
} catch (e) {
|
||
console.error(e);
|
||
setPill($("sessionPill"), "Session failed", "err");
|
||
alert("Start session failed: " + e.message);
|
||
} finally {
|
||
// allow a new poll loop to start later
|
||
trainingPollAbort = false;
|
||
}
|
||
});
|
||
|
||
$("resetBtn").addEventListener("click", async () => {
|
||
try {
|
||
await api("/api/reset_recordings", {method:"POST"});
|
||
currentSpeaker = 1;
|
||
currentTake = 0;
|
||
$("takesList").textContent = "";
|
||
refreshUI();
|
||
setPill($("status"), "Recordings reset", "ok");
|
||
} catch (e) {
|
||
alert("Reset failed: " + e.message);
|
||
}
|
||
});
|
||
|
||
$("beginBtn").addEventListener("click", async () => {
|
||
if (!session) { alert("Start a session first."); return; }
|
||
try {
|
||
await ensureMic();
|
||
} catch (e) {
|
||
alert("Mic permission failed: " + e.message);
|
||
return;
|
||
}
|
||
|
||
$("takesList").textContent = "";
|
||
refreshUI();
|
||
|
||
isRunning = true;
|
||
$("beginBtn").disabled = true;
|
||
|
||
setPill($("speakerState"), `Speaker ${currentSpeaker}/${speakersTotal}`);
|
||
setPill($("status"), "Listening… say the wake word now", "ok");
|
||
setPill($("takeState"), "Listening…", "ok");
|
||
});
|
||
</script>
|
||
</body>
|
||
</html> |