Skip to content
AI

Podcast Auto-Chapters and Show Notes

Drop an episode and get timestamps, chapter titles and show notes.

Intermediate16h1–3 people16/20 fit
TL;DR

Build Podcast Auto-Chapters and Show Notes for indie podcasters and youtubers in ~16h with 1–3 people. Stack: Whisper + Embeddings + LLM + Next.js.

Why it wins: Saves creators real hours every week and the timestamps are easy to verify live.

Build flow

01 · Problem
Writing chapters and show notes for a one-hour episode takes creators an hour or more every week.
02 · Solution
Transcribe with Whisper, detect topic shifts with embeddings over transcript windows, then have an LLM title each chapter and draft show notes with links mentioned.
03 · Build
  1. 1Transcribe audio with Whisper timestamps
  2. 2Embed 60-second windows and find similarity drops
  3. 3Ask the LLM to title each segment
  4. 4Format as YouTube chapters and Markdown notes
04 · Demo
Play the episode and click a generated chapter to jump exactly to the topic change.

MVP checklist

  • Chapter timestamps
  • Show notes draft
  • Mentioned links list
  • YouTube description export

Tech stack

TechRole
WhisperSpeech-to-text
EmbeddingsSemantic similarity
LLMLanguage model
Next.jsFrontend + API routes

Reference repos

Open-source projects you can build on. Links checked 2026-10-31.

Related ideas

More AI