sqwartHome

Podcast Show Notes Generator

Turn an episode into show notes, with a summary, key points and chapters, without uploading it.

  • Free
  • No upload
  • No watermark
  • No sign-up
  • AI runs on your device
How it works

How to use the podcast show notes generator

  1. 1

    Choose the episode's audio or video, or its SRT or VTT transcript.

  2. 2

    Pick the language and how many chapters you want.

  3. 3

    Press Write show notes and let it listen.

  4. 4

    Edit, then copy the notes into your podcast host or YouTube description.

In detail

What the show notes contain

  • A short summary of the episode, to open the notes and to use in social posts.
  • In this episode: the main points, as a list listeners can skim.
  • Chapters: timestamps with titles, which Spotify, Apple Podcasts (via your host) and YouTube can turn into chapters.
  • The full transcript as an SRT file, if you want to publish it or caption a video version.

How the chapters are found

First the speech becomes a timed transcript with Whisper, running in your browser (or, if you already have one, open the SRT or VTT file and this step is skipped). Then the tool looks for the moments the topic changes: at every pause it compares the words of the minute before with the words of the minute after, and the places where they share the least become chapter starts, kept far enough apart that each chapter is a real section rather than a sentence. A 20-minute video gets about eight chapters by default; you can choose from 3 to 12.

In Chrome on a computer, its built-in AI (Gemini Nano, on your device) reads each chapter and gives it a short title. In other browsers each title is the chapter's most distinctive phrase (the words that come up in it and not elsewhere), which is accurate but plainer, so you may want to polish the wording. Nothing is uploaded in either case.

Summaries with and without Chrome's AI

In Chrome on a computer, the summary and key points are written by its built-in AI, on your device. In other browsers the tool picks the most central sentences of the transcript instead: accurate, because they are the speakers' own words, but less polished, so treat them as a draft. Either way, read the notes before publishing: AI can misstate a detail, and spoken sentences don't always read well.

FAQ

Questions, answered.

Is my episode uploaded?

No. The audio is transcribed in your browser and the notes are written on your device.

How long does a one-hour episode take?

With the Balanced model, roughly 15 to 30 minutes on a laptop processor, much less with WebGPU. With an existing SRT transcript it takes seconds.

Can it tell the hosts apart?

No. Whisper doesn't label speakers, so the notes describe what is said, not who said it.

Does it support Hindi podcasts?

Yes for the transcript and chapters (choose Hindi and the Accurate model). Chrome's built-in AI writes summaries in English.

Is there a length limit?

No set limit. Very long episodes need more memory and time; a laptop handles a couple of hours.

Keep going

More free creator tools