From 0f24cf5ec91474456d9989c10b2cea1a584f66ac Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E2=80=9CNaeel=E2=80=9D?= <tazet@narod.ru>
Date: Sat, 23 May 2026 06:13:46 +0400
Subject: [PATCH] =?UTF-8?q?v80:=20=D0=BE=D0=B1=D1=80=D0=B5=D0=B7=D0=BA?=
 =?UTF-8?q?=D0=B0=200.2=D1=81=20=D0=B2=20=D0=BD=D0=B0=D1=87=D0=B0=D0=BB?=
 =?UTF-8?q?=D0=B5=20=D0=B7=D0=B0=D0=BF=D0=B8=D1=81=D0=B8=20=E2=80=94=20?=
 =?UTF-8?q?=D1=83=D0=B1=D0=B8=D1=80=D0=B0=D0=B5=D1=82=20=D1=89=D0=B5=D0=BB?=
 =?UTF-8?q?=D1=87=D0=BE=D0=BA=20(=D0=B2=D0=BC=D0=B5=D1=81=D1=82=D0=BE=20tr?=
 =?UTF-8?q?imSilence)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 index.html       |  2 +-
 js/audio.js      | 50 ++++++++++++++++++++++++++++++++++++------------
 js/main.js       |  2 +-
 js/transcribe.js |  7 +++++--
 4 files changed, 45 insertions(+), 16 deletions(-)

diff --git a/index.html b/index.html
index 7416e03..6f42bfc 100644
--- a/index.html
+++ b/index.html
@@ -13,7 +13,7 @@
 <div id="app">
   <h1>🇮🇹 Lyngvo <span style="font-size:0.5em;color:#aaa" id="ver"></span> <span style="font-size:0.35em;color:#999;font-weight:normal">тренажёр итальянского произношения · для Google Chrome</span></h1>
   <div style="font-size:0.8em;color:#666;margin-bottom:14px;white-space:nowrap">
-    1. Введите фразу &nbsp;→&nbsp; 2. ▶ Эталон (прослушайте) &nbsp;→&nbsp; 3. 🎙 Записать (произнесите) &nbsp;→&nbsp; 4. 📊 Сравнить
+    1. Введите фразу &nbsp;→&nbsp; 2. ▶ Эталон (прослушайте) &nbsp;→&nbsp; 3. 🎙 Записать (сделайте короткую паузу перед произнесением) &nbsp;→&nbsp; 4. 📊 Сравнить
   </div>
   <textarea id="inputText" rows="3" placeholder="Введите итальянское слово или фразу..."></textarea>
   <div id="history"></div>
diff --git a/js/audio.js b/js/audio.js
index 590c77b..6661997 100644
--- a/js/audio.js
+++ b/js/audio.js
@@ -90,23 +90,20 @@ function stopRecording() {
   if (mediaRecorder && mediaRecorder.state === 'recording') mediaRecorder.stop();
 }
 
-// ---------- Воспроизведение (с обрезкой тишины) ----------
+// ---------- Воспроизведение (с обрезкой 0.2с в начале) ----------
 async function playTrimmed(blob) {
-  const t = await trimSilence(blob);
   const ctx = await getAudioCtx();
-  let buf;
-  if (t) {
-    const offline = new OfflineAudioContext(1, t.data.length, t.sr);
-    buf = offline.createBuffer(1, t.data.length, t.sr);
-    buf.getChannelData(0).set(t.data);
-  } else {
-    const ab = await blob.arrayBuffer();
-    buf = await ctx.decodeAudioData(ab);
-  }
+  const ab = await blob.arrayBuffer();
+  const rawBuf = await ctx.decodeAudioData(ab);
+  const sr = rawBuf.sampleRate;
+  const data = rawBuf.getChannelData(0);
+  const cutSamples = Math.floor(sr * 0.2);
+  const trimmed = data.slice(cutSamples);
+  const buf = ctx.createBuffer(1, trimmed.length, sr);
+  buf.getChannelData(0).set(trimmed);
   const src = ctx.createBufferSource();
   src.buffer = buf;
   const gain = ctx.createGain();
-  // Fade-in 10мс — убирает щелчок при старте
   gain.gain.setValueAtTime(0, ctx.currentTime);
   gain.gain.linearRampToValueAtTime(1, ctx.currentTime + 0.01);
   src.connect(gain).connect(ctx.destination);
@@ -305,3 +302,32 @@ async function trimSilence(blob) {
     return { buffer: trimmedBuf, data: data.slice(startSample, endSample), sr, startSample, endSample };
   } catch(e) { return null; }
 }
+
+// ---------- WAV кодирование (для отправки в Whisper) ----------
+function encodeWAV(samples, sampleRate) {
+  const buf = new ArrayBuffer(44 + samples.length * 2);
+  const v = new DataView(buf);
+  const w = (o, s) => { for (let i=0;i<s.length;i++) v.setUint8(o+i, s.charCodeAt(i)); };
+  w(0, 'RIFF'); v.setUint32(4, 36+samples.length*2, true); w(8, 'WAVE');
+  w(12, 'fmt '); v.setUint32(16, 16, true); v.setUint16(20, 1, true); v.setUint16(22, 1, true);
+  v.setUint32(24, sampleRate, true); v.setUint32(28, sampleRate*2, true); v.setUint16(32, 2, true); v.setUint16(34, 16, true);
+  w(36, 'data'); v.setUint32(40, samples.length*2, true);
+  for (let i=0;i<samples.length;i++) {
+    const s = Math.max(-1, Math.min(1, samples[i]));
+    v.setInt16(44+i*2, s<0?s*0x8000:s*0x7FFF, true);
+  }
+  return buf;
+}
+
+// Обрезает первые cutSec секунд, возвращает WAV Blob (для Groq)
+async function trimStartBlob(blob, cutSec = 0.2) {
+  const ctx = new OfflineAudioContext(1, 1, 44100);
+  const ab = await blob.arrayBuffer();
+  const buf = await ctx.decodeAudioData(ab);
+  const sr = buf.sampleRate;
+  const data = buf.getChannelData(0);
+  const cutSamples = Math.floor(sr * cutSec);
+  const trimmed = data.slice(cutSamples);
+  const wav = encodeWAV(trimmed, sr);
+  return new Blob([wav], { type: 'audio/wav' });
+}
diff --git a/js/main.js b/js/main.js
index a35ad42..90fd8e0 100644
--- a/js/main.js
+++ b/js/main.js
@@ -1,6 +1,6 @@
 // ---------- Глобальные переменные и инициализация ----------
 const GROQ_API_KEY = '';
-const VERSION = 'v79';
+const VERSION = 'v80';
 
 let mediaRecorder, audioChunks = [], userAudioBlob = null, ttsText = '';
 let isRecording = false, isAnalyzing = false, micAnalyser = null, micAnimId = null;
diff --git a/js/transcribe.js b/js/transcribe.js
index f5313db..3ffbfe3 100644
--- a/js/transcribe.js
+++ b/js/transcribe.js
@@ -4,12 +4,15 @@ async function transcribe(blob) {
   if (!GROQ_API_KEY) return { text: '', error: 'no_key' };
   const tStart = performance.now();
 
+  // Обрезаем 0.2с в начале (щелчок) — декодируем → WAV
+  const trimmedBlob = await trimStartBlob(blob);
+
   const form = new FormData();
-  form.append('file', blob, 'audio.webm');
+  form.append('file', trimmedBlob, 'audio.wav');
   form.append('model', 'whisper-large-v3');
   form.append('language', 'it');
 
-  console.log('[HTTP] blob ' + (blob.size/1024).toFixed(1) + 'KB → multipart POST');
+  console.log('[HTTP] blob ' + (trimmedBlob.size/1024).toFixed(1) + 'KB → multipart POST');
 
   try {
     const r = await fetch('https://lang.kube5s.ru/openai/v1/audio/transcriptions', {