Recipe

Auto-generate YouTube chapters from transcript

Turn raw video transcripts into timestamped chapter markers using Meridian's structured output pipeline.

Ingredients

A YouTube video URL with captions available
Meridian API key with chapters scope
Python 3.10+ or cURL

Step 1 — Fetch the transcript

Use yt-dlp or the YouTube Data API to pull the timed caption track. Meridian accepts SRT, VTT, or plain text with inline timestamps.

Step 2 — Send to Meridian

POST the transcript to the chapters endpoint with a schema describing your desired chapter density.

curl -X POST https://api.getnimbus.net/v1/chapters \
  -H "Authorization: Bearer $MERIDIAN_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "transcript": "<srt content>",
    "max_chapters": 8,
    "min_segment_sec": 30
  }'

Step 3 — Parse the response

Meridian returns a JSON array of chapter objects with start_sec, title, and an optional summary. Paste the timestamps directly into your YouTube description or video editor.

Pro tip

Set min_segment_sec to at least 30 to avoid noisy chapter boundaries. Combine with Meridian's summarization endpoint for chapter descriptions that boost SEO.

← Back to all recipes