Transcript API for YouTube videos
Pull spoken text from any public YouTube video with a single REST call. JSON segments with timestamps, language selection, and no scraper maintenance.
- Structured JSON with start time and duration per line
- List available caption languages before fetching
- 20 free requests per hour without an account
A transcript API turns a YouTube URL into structured text you can store, search, or feed into an LLM. Instead of parsing watch pages or running headless browsers, you call one HTTP endpoint and receive JSON with optional per-line timestamps. FreeTranscriptAPI hosts that service at api.freetranscriptapi.com/v1.
What “transcript API” means in practice
You send a video identifier, the service returns caption text YouTube already publishes for that video. The API does not download video files, does not bypass sign-in walls, and is not affiliated with Google. It is built for developers who need repeatable caption fetches in production jobs, dashboards, and automation.
The primary endpoint is GET /transcript with a video_url query parameter. You can pass a full watch URL, a short youtu.be link, or an 11-character video ID. Optional lang selects a caption language; optional format switches between JSON and plain text.
JSON shape and why timestamps matter
Default JSON includes the resolved language, video title, and a transcript array. Each item has text, start (seconds from the beginning), and duration. That structure maps cleanly to subtitle players, chapter generators, and citation UIs (“see 12:04 in the video”).
If you only need a blob of text for summarization, pass format=text. Keep JSON when you sync captions to a video player or split long videos into time-based chunks before sending them to a model.
List languages before you fetch
Videos often ship with multiple tracks: manual translations, auto-generated English, or regional variants. Calling GET /languages first avoids guessing. The response lists language codes, human-readable names, and whether each track was auto-generated. You then pass the code you want on /transcript. When a requested language is missing, the API falls back according to its documented rules instead of failing silently with the wrong locale.
Transcript API vs scraping
Scraping looks free until YouTube changes markup, rate-limits your IPs, or blocks cloud provider ranges. npm packages that mimic the watch page face the same maintenance cycle. A hosted transcript API centralizes that work: you get stable URLs, published rate limits, and support when caption delivery changes.
Choose scraping for one-off scripts on your laptop. Choose an API when transcripts sit on your critical path: nightly ETL, customer-facing search, or compliance archives. Read the without-scraping guide for a longer comparison.
Try it without an API key
Anonymous callers get 20 requests per hour per IP. That is enough to prototype parsers, validate JSON shape, and run integration tests in CI with modest volume. When you outgrow the limit, create a free account for 1,000 signup credits and a bearer token. No credit card is required for signup.
Example request
curl "https://api.freetranscriptapi.com/v1/transcript?video_url=https://www.youtube.com/watch?v=dQw4w9WgXcQ"Where teams use a YouTube transcript API
- Search indexes that need spoken content from public marketing videos
- LLM pipelines that summarize webinars or podcast episodes on YouTube
- Accessibility workflows that export captions for review or translation
- Competitive intelligence tools that track messaging across channels
- n8n or Zapier flows that trigger when a new video URL appears in a sheet
For endpoint details, authentication headers, error codes, and MCP integration, see the API reference. For a walkthrough aimed at new integrators, start with the developer guide.
Anonymous callers get 20 requests per hour per IP without an API key.
More guides
Other topics on the same API: