Transcript — Getting Started

What is the transcript module?

Lyra SDK includes a built-in module for fetching YouTube video transcripts. Unlike the core SDK (which uses the YouTube Data API v3), the transcript module accesses YouTube's internal Innertube API directly.

Key differences from the core SDK:

Feature	Core SDK (`yt()`)	Transcript module
API key required	Yes	No
Quota consumption	Yes	No
Import path	`lyra-sdk`	`lyra-sdk/transcript`
Data source	YouTube Data API v3	YouTube Innertube API
What you get	Video/channel/playlist info	Captions/transcript text

Installation

The transcript module is included in lyra-sdk — no separate install needed:

npm install lyra-sdk

pnpm add lyra-sdk

yarn add lyra-sdk

bun add lyra-sdk

Your first transcript

import { transcribeVideo } from 'lyra-sdk/transcript'

const lines = await transcribeVideo('dQw4w9WgXcQ')

for (const line of lines) {
  console.log(`[${line.offset}s] ${line.text}`)
}

No API key, no yt() client, no configuration needed. Just import and call.

Fetch a specific language

const lines = await transcribeVideo('dQw4w9WgXcQ', { lang: 'es' })

To see what languages are available:

import { listCaptionTracks } from 'lyra-sdk/transcript'

const tracks = await listCaptionTracks('dQw4w9WgXcQ')
// [{ languageCode: 'en', languageName: 'English', isAutoGenerated: true }, ...]

Get video metadata alongside transcript

const result = await transcribeVideo('dQw4w9WgXcQ', { includeMeta: true })

console.log(result.meta.title)   // "Rick Astley - Never Gonna Give You Up"
console.log(result.meta.author)  // "Rick Astley"
console.log(result.meta.views)   // 1600000000
console.log(result.lines.length) // 61

Output as SRT or VTT

import { transcribeVideo, toSRT, toVTT, toPlainText } from 'lyra-sdk/transcript'

const lines = await transcribeVideo('dQw4w9WgXcQ')

// SubRip format (.srt)
const srt = toSRT(lines)

// WebVTT format (.vtt)
const vtt = toVTT(lines)

// Plain text with custom separator
const text = toPlainText(lines, ' ')

Custom networking

Custom User-Agent

const lines = await transcribeVideo('dQw4w9WgXcQ', {
  userAgent: 'MyApp/1.0',
})

HTTP mode (no TLS)

const lines = await transcribeVideo('dQw4w9WgXcQ', {
  useHttp: true,
})

useHttp: true disables HTTPS. Not recommended for production use.

Proxy via custom fetch

const lines = await transcribeVideo('dQw4w9WgXcQ', {
  customFetch: async (url, init) => {
    return fetch(`https://my-proxy.com/fetch?url=${encodeURIComponent(url)}`, init)
  },
})

AbortSignal

const controller = new AbortController()
setTimeout(() => controller.abort(), 10000) // 10s timeout

const lines = await transcribeVideo('dQw4w9WgXcQ', {
  signal: controller.signal,
})

Reusable client

For multiple requests with shared configuration, use TranscriptClient:

import { TranscriptClient, InMemoryCache } from 'lyra-sdk/transcript'

const client = new TranscriptClient({
  lang: 'en',
  retries: 2,
  retryDelay: 500,
  cache: new InMemoryCache(),
})

// All calls inherit the config above
const lines1 = await client.transcribe('dQw4w9WgXcQ')       // cold fetch + cached
const lines2 = await client.transcribe('dQw4w9WgXcQ')       // cache hit
const tracks = await client.availableTracks('jNQXAC9IVRw')   // new video