語音辨識

// 最佳音訊設定
const optimalConfig = {
  sampleRate: 16000,    // 16kHz 採樣率
  encoding: 'LINEAR16', // 線性 PCM 編碼
  channels: 1,          // 單聲道
  bitDepth: 16         // 16-bit 位元深度
};

const result = await client.transcribe({
  audio: audioBuffer,
  language: 'mixed',
  ...optimalConfig
});

const stream = client.createRealTimeStream({
  language: 'mixed',
  sampleRate: 16000,
  encoding: 'LINEAR16',
  interimResults: true // 顯示中間結果
});

stream.on('data', (result) => {
  if (result.isFinal) {
    console.log('最終結果:', result.transcript);
  } else {
    console.log('中間結果:', result.transcript);
  }
});

const result = await client.transcribe({
  audio: fs.readFileSync('recording.wav'),
  language: 'mixed',
  enableWordTimeOffsets: true, // 啟用時間軸
  enableSpeakerDiarization: true // 說話者分離
});

console.log('完整逐字稿:', result.transcript);

// 顯示時間軸資訊
result.words.forEach(word => {
  console.log(`${word.text} (${word.startTime}s - ${word.endTime}s)`);
});

const result = await client.transcribe({
  audio: audioBuffer,
  language: 'mixed',
  enableSpeakerDiarization: true,
  maxSpeakers: 4 // 最多 4 位說話者
});

// 顯示說話者資訊
result.segments.forEach(segment => {
  console.log(`說話者 ${segment.speaker}: ${segment.text}`);
  console.log(`時間: ${segment.startTime}s - ${segment.endTime}s`);
});

const result = await client.transcribe({
  audio: audioBuffer,
  language: 'mixed',
  enableWordConfidence: true
});

// 顯示低信心度的詞彙
result.words.forEach(word => {
  if (word.confidence < 0.8) {
    console.log(`低信心度詞彙: ${word.text} (${word.confidence})`);
  }
});

// 音訊預處理建議
const preprocessAudio = (audioBuffer) => {
  // 1. 降噪處理
  const denoisedAudio = removeNoise(audioBuffer);
  
  // 2. 音量正規化
  const normalizedAudio = normalizeVolume(denoisedAudio);
  
  // 3. 移除靜音段落
  const trimmedAudio = trimSilence(normalizedAudio);
  
  return trimmedAudio;
};

const result = await client.transcribe({
  audio: preprocessAudio(rawAudio),
  language: 'mixed'
});

// 大檔案分段處理
const processLargeFile = async (largeAudioFile) => {
  const chunks = splitAudioIntoChunks(largeAudioFile, 60); // 60秒分段
  const results = [];
  
  for (const chunk of chunks) {
    const result = await client.transcribe({
      audio: chunk,
      language: 'mixed'
    });
    results.push(result);
  }
  
  return mergeResults(results);
};

try {
  const result = await client.transcribe({
    audio: audioBuffer,
    language: 'mixed'
  });
} catch (error) {
  switch (error.code) {
    case 'AUDIO_TOO_SHORT':
      console.error('音訊檔案太短，至少需要 1 秒');
      break;
    case 'AUDIO_TOO_LONG':
      console.error('音訊檔案太長，最多 60 分鐘');
      break;
    case 'UNSUPPORTED_FORMAT':
      console.error('不支援的音訊格式');
      break;
    case 'POOR_AUDIO_QUALITY':
      console.error('音訊品質過低，建議重新錄製');
      break;
    case 'NO_SPEECH_DETECTED':
      console.error('未偵測到語音內容');
      break;
    default:
      console.error('辨識錯誤:', error.message);
  }
}