v73: обрезка тишины (начало/конец) → WAV

dist/index.html

@@ -257,7 +257,7 @@ async function transcribe(blob) {
   const tStart = performance.now();
 
   const form = new FormData();
-  form.append('file', blob, 'audio.webm');
+  form.append('file', blob, blob.type === 'audio/wav' ? 'audio.wav' : 'audio.webm');
   form.append('model', 'whisper-large-v3');
   form.append('language', 'it');
 
@@ -365,11 +365,15 @@ async function startRecording() {
   const opts = mimeType ? { mimeType } : {};
   mediaRecorder = new MediaRecorder(stream, opts);
   mediaRecorder.ondataavailable = e => { if (e.data.size > 0) audioChunks.push(e.data); };
-  mediaRecorder.onstop = () => {
+  mediaRecorder.onstop = async () => {
     stopMicMeter();
     recordDuration = (Date.now() - recordStartTime) / 1000;
-    userAudioBlob = new Blob(audioChunks, { type: mimeType || 'audio/webm' });
+    let blob = new Blob(audioChunks, { type: mimeType || 'audio/webm' });
     stream.getTracks().forEach(t => t.stop());
+    // Обрезаем тишину
+    const trimmed = await trimSilence(blob);
+    if (trimmed && trimmed.size > 0) blob = trimmed;
+    userAudioBlob = blob;
     enable('btnCompare', true);
     enable('btnStereo', true);
     enable('btnPlayUser', true);
@@ -554,6 +558,74 @@ function pitchFrame(data, start, len, sr) {
   return bestR > 0.30 ? sr / bestLag : 0;
 }
 
+// ---------- Обрезка тишины (щелчок в начале, тишина в конце) ----------
+async function trimSilence(blob) {
+  try {
+    const ctx = new (window.OfflineAudioContext || window.webkitOfflineAudioContext)(1, 1, 44100);
+    const ab = await blob.arrayBuffer();
+    const buf = await ctx.decodeAudioData(ab);
+    const data = buf.getChannelData(0);
+    const sr = buf.sampleRate;
+    const len = data.length;
+
+    // RMS по окнам 10мс
+    const win = Math.floor(sr * 0.01);
+    const rms = [];
+    let maxRms = 0;
+    for (let i = 0; i < len; i += win) {
+      let sum = 0, n = 0;
+      for (let j = i; j < i + win && j < len; j++, n++) sum += data[j] * data[j];
+      const v = Math.sqrt(sum / n);
+      rms.push(v);
+      if (v > maxRms) maxRms = v;
+    }
+    if (maxRms < 0.001) return blob; // тишина
+
+    const threshold = maxRms * 0.02; // 2% от пика
+
+    // Начало — первый кадр выше порога
+    let startIdx = 0;
+    for (let i = 0; i < rms.length; i++) {
+      if (rms[i] > threshold) { startIdx = Math.max(0, i - 2); break; }
+    }
+
+    // Конец — последний кадр выше порога
+    let endIdx = rms.length - 1;
+    for (let i = rms.length - 1; i >= 0; i--) {
+      if (rms[i] > threshold) { endIdx = Math.min(rms.length - 1, i + 2); break; }
+    }
+
+    const startSample = startIdx * win;
+    const endSample = Math.min((endIdx + 1) * win, len);
+    if (endSample - startSample < sr * 0.08) return blob; // слишком коротко
+
+    // WAV: 44-byte header + PCM16
+    const trimmed = data.slice(startSample, endSample);
+    const numSamples = trimmed.length;
+    const wav = new ArrayBuffer(44 + numSamples * 2);
+    const view = new DataView(wav);
+    const writeStr = (off, s) => { for (let i = 0; i < s.length; i++) view.setUint8(off + i, s.charCodeAt(i)); };
+    writeStr(0, 'RIFF');
+    view.setUint32(4, 36 + numSamples * 2, true);
+    writeStr(8, 'WAVE');
+    writeStr(12, 'fmt ');
+    view.setUint32(16, 16, true);
+    view.setUint16(20, 1, true);
+    view.setUint16(22, 1, true);
+    view.setUint32(24, sr, true);
+    view.setUint32(28, sr * 2, true);
+    view.setUint16(32, 2, true);
+    view.setUint16(34, 16, true);
+    writeStr(36, 'data');
+    view.setUint32(40, numSamples * 2, true);
+    for (let i = 0; i < numSamples; i++) {
+      const s = Math.max(-1, Math.min(1, trimmed[i]));
+      view.setInt16(44 + i * 2, s < 0 ? s * 0x8000 : s * 0x7FFF, true);
+    }
+    return new Blob([wav], { type: 'audio/wav' });
+  } catch(e) { return blob; }
+}
+
 </script>
 <script>
 // ---------- Сравнение текстов (Левенштейн) и doCompare ----------
@@ -694,7 +766,7 @@ async function doCompare(blob, originalText, duration) {
 <script>
 // ---------- Глобальные переменные и инициализация ----------
 const GROQ_API_KEY = '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';
-const VERSION = 'v72';
+const VERSION = 'v73';
 
 let mediaRecorder, audioChunks = [], userAudioBlob = null, ttsText = '';
 let isRecording = false, isAnalyzing = false, micAnalyser = null, micAnimId = null;

js/audio.js

@@ -62,11 +62,15 @@ async function startRecording() {
   const opts = mimeType ? { mimeType } : {};
   mediaRecorder = new MediaRecorder(stream, opts);
   mediaRecorder.ondataavailable = e => { if (e.data.size > 0) audioChunks.push(e.data); };
-  mediaRecorder.onstop = () => {
+  mediaRecorder.onstop = async () => {
     stopMicMeter();
     recordDuration = (Date.now() - recordStartTime) / 1000;
-    userAudioBlob = new Blob(audioChunks, { type: mimeType || 'audio/webm' });
+    let blob = new Blob(audioChunks, { type: mimeType || 'audio/webm' });
     stream.getTracks().forEach(t => t.stop());
+    // Обрезаем тишину
+    const trimmed = await trimSilence(blob);
+    if (trimmed && trimmed.size > 0) blob = trimmed;
+    userAudioBlob = blob;
     enable('btnCompare', true);
     enable('btnStereo', true);
     enable('btnPlayUser', true);
@@ -250,3 +254,71 @@ function pitchFrame(data, start, len, sr) {
   }
   return bestR > 0.30 ? sr / bestLag : 0;
 }
+
+// ---------- Обрезка тишины (щелчок в начале, тишина в конце) ----------
+async function trimSilence(blob) {
+  try {
+    const ctx = new (window.OfflineAudioContext || window.webkitOfflineAudioContext)(1, 1, 44100);
+    const ab = await blob.arrayBuffer();
+    const buf = await ctx.decodeAudioData(ab);
+    const data = buf.getChannelData(0);
+    const sr = buf.sampleRate;
+    const len = data.length;
+
+    // RMS по окнам 10мс
+    const win = Math.floor(sr * 0.01);
+    const rms = [];
+    let maxRms = 0;
+    for (let i = 0; i < len; i += win) {
+      let sum = 0, n = 0;
+      for (let j = i; j < i + win && j < len; j++, n++) sum += data[j] * data[j];
+      const v = Math.sqrt(sum / n);
+      rms.push(v);
+      if (v > maxRms) maxRms = v;
+    }
+    if (maxRms < 0.001) return blob; // тишина
+
+    const threshold = maxRms * 0.02; // 2% от пика
+
+    // Начало — первый кадр выше порога
+    let startIdx = 0;
+    for (let i = 0; i < rms.length; i++) {
+      if (rms[i] > threshold) { startIdx = Math.max(0, i - 2); break; }
+    }
+
+    // Конец — последний кадр выше порога
+    let endIdx = rms.length - 1;
+    for (let i = rms.length - 1; i >= 0; i--) {
+      if (rms[i] > threshold) { endIdx = Math.min(rms.length - 1, i + 2); break; }
+    }
+
+    const startSample = startIdx * win;
+    const endSample = Math.min((endIdx + 1) * win, len);
+    if (endSample - startSample < sr * 0.08) return blob; // слишком коротко
+
+    // WAV: 44-byte header + PCM16
+    const trimmed = data.slice(startSample, endSample);
+    const numSamples = trimmed.length;
+    const wav = new ArrayBuffer(44 + numSamples * 2);
+    const view = new DataView(wav);
+    const writeStr = (off, s) => { for (let i = 0; i < s.length; i++) view.setUint8(off + i, s.charCodeAt(i)); };
+    writeStr(0, 'RIFF');
+    view.setUint32(4, 36 + numSamples * 2, true);
+    writeStr(8, 'WAVE');
+    writeStr(12, 'fmt ');
+    view.setUint32(16, 16, true);
+    view.setUint16(20, 1, true);
+    view.setUint16(22, 1, true);
+    view.setUint32(24, sr, true);
+    view.setUint32(28, sr * 2, true);
+    view.setUint16(32, 2, true);
+    view.setUint16(34, 16, true);
+    writeStr(36, 'data');
+    view.setUint32(40, numSamples * 2, true);
+    for (let i = 0; i < numSamples; i++) {
+      const s = Math.max(-1, Math.min(1, trimmed[i]));
+      view.setInt16(44 + i * 2, s < 0 ? s * 0x8000 : s * 0x7FFF, true);
+    }
+    return new Blob([wav], { type: 'audio/wav' });
+  } catch(e) { return blob; }
+}

js/main.js

@@ -1,6 +1,6 @@
 // ---------- Глобальные переменные и инициализация ----------
 const GROQ_API_KEY = '';
-const VERSION = 'v72';
+const VERSION = 'v73';
 
 let mediaRecorder, audioChunks = [], userAudioBlob = null, ttsText = '';
 let isRecording = false, isAnalyzing = false, micAnalyser = null, micAnimId = null;

js/transcribe.js

@@ -5,7 +5,7 @@ async function transcribe(blob) {
   const tStart = performance.now();
 
   const form = new FormData();
-  form.append('file', blob, 'audio.webm');
+  form.append('file', blob, blob.type === 'audio/wav' ? 'audio.wav' : 'audio.webm');
   form.append('model', 'whisper-large-v3');
   form.append('language', 'it');