@lcharleslaing/transcripts

Transcript and song-lyric parsing utilities for SRT, VTT, LRC, TXT, JSON, and Whisper Music phrase JSON.

Supported formats

• SRT
• WebVTT / VTT
• LRC
• TXT
• Generic JSON segment/chunk formats
• Whisper Music phrase JSON, including .phrase.json and .phrases.json

All formats normalize to the same structure:

interface TranscriptLine {
  start: number | null;
  end: number | null;
  text: string;
}

interface Transcript {
  format: TranscriptFormat;
  lines: TranscriptLine[];
  text: string;
}

Phrase-level Whisper Music chunks stay phrase-level; they are not merged into larger subtitle blocks.

Usage

import { detectTranscriptFormat, parseTranscript } from '@lcharleslaing/transcripts';

const content = '[00:01.20]First lyric line\n[00:03.40]Second lyric line';

const format = detectTranscriptFormat(content, 'song.lrc');
const transcript = parseTranscript(content, { filename: 'song.lrc' });

console.log(format); // lrc
console.log(transcript.lines);

You can force a known format:

const transcript = parseTranscript(jsonContent, {
  format: 'phrase-json'
});

JSON compatibility

The JSON parser recognizes common collections such as:

• segments
• lines
• items
• transcript
• phrases
• phrase_segments
• phraseSegments
• chunks

Common text fields include text, phrase, content, line, lyrics, and lyric.

Common timing fields include start, start_time, startTime, end, end_time, endTime, duration, and related aliases.

Development

npm install
npm start

Quality checks

npm run check

Build

npm run build

License

MIT