@etaio/audio-extraction

Audio extraction and transcription services for the Etaio platform.

Features

• Audio Extraction: Extract audio from video files using FFmpeg
• Whisper Transcription: Transcribe audio using OpenAI Whisper API
• Word-Level Timestamps: Get precise word-level timing information
• Multiple Audio Formats: Support for MP3, WAV, AAC, M4A, FLAC, OGG
• Cost Tracking: Calculate transcription costs automatically

Installation

pnpm install @etaio/audio-extraction

Usage

Extract Audio from Video

import { extractAudio } from '@etaio/audio-extraction';

const audioInfo = await extractAudio('/path/to/video.mp4', {
  format: 'mp3',
  bitrate: 192,
  sampleRate: 48000,
  channels: 2
});

console.log('Audio extracted:', audioInfo);

Transcribe Audio

import { transcribeAudio } from '@etaio/audio-extraction';

const transcript = await transcribeAudio('/path/to/audio.mp3', {
  apiKey: 'your-openai-api-key',
  model: 'whisper-1',
  language: 'en',
  wordTimestamps: true,
  temperature: 0.0
});

console.log('Transcript:', transcript.fullText);
console.log('Segments:', transcript.segments);

Process Transcript

import { processTranscriptSegments, calculateTranscriptStatistics } from '@etaio/audio-extraction';

const segments = processTranscriptSegments(transcript);
const stats = calculateTranscriptStatistics(transcript);

console.log('Word count:', stats.wordCount);
console.log('Duration:', stats.duration);

API Reference

See TypeScript definitions for complete API documentation.

License

MIT