YouTube Transcript API in Python: A 5-Minute Quickstart

One request, one transcript

The shortest path is a single GET to /v1/transcripts/{videoId} with your key in the Authorization header. The response is JSON: video id, language, duration and a list of segments. No SDK is required — Python’s requests or httpx is enough.

Choosing a language and format

Call /v1/transcripts/{videoId}/languages first when a video may be multilingual, pick the track you want, then pass ?lang= and ?format=. Asking for srt or vtt returns ready-to-save subtitles; asking for json keeps the structured segments.

Batching and retries

For lists of videos, POST up to 50 ids to /v1/transcripts in one call and read per-item results. Wrap requests with a timeout and a small exponential backoff: transient upstream hiccups and 429 rate limits both clear quickly, and every response carries an X-Request-ID you can log for support.

Handling the awkward cases

Some videos have no captions, some are private, and some are region-locked. The API answers with a typed error code and a human-readable message, so your code branches on the code rather than parsing prose. Log the request id; that is the fastest route to a fix.