Writing Show Notes, Chapters and Summaries from an Episode Transcript

What good show notes contain

Show notes do several jobs at once: they persuade someone to press play, help a listener find a moment again, credit guests and sources, and give search engines and podcast apps text to work with. A useful set usually has:

Summary

Two to four sentences on what the listener will get from the episode, not a list of everything discussed.

Chapters

Timestamps with short labels, so listeners can skip to a topic.

Guest

Name, role and where to find them, spelled exactly as they spell it.

Links and resources

Every book, tool, article or person mentioned that a listener might look up.

Quotes

One or two lines worth sharing, checked against the audio.

Transcript

A link to the full text, for accessibility and search.

The transcript feeds every row. Without it, writing these means scrubbing back through the audio for each detail.

Get a transcript of the final mix

1. Export the episode exactly as published, after edits, intro and ad reads, so the timestamps match what listeners hear.
2. Upload it to audio to text as MP3, WAV, M4A, AAC, OGG or FLAC, in transcript mode.
3. Download the timestamped transcript for chapters and quotes and the plain transcript for searching.
4. Search for the guest's name, company names and any terms the speakers spelled out, and fix them first.

A transcript costs 1 credit per minute, so a 52-minute episode is 52 credits, about 5 cents. Transcribing a pre-edit version is a common mistake: every timestamp drifts after the first cut.

Building chapters from timestamps

Read the timestamped transcript quickly and mark the lines where the conversation turns to a new subject. Questions from the host are reliable markers in interview shows. Aim for chapters that are long enough to be worth jumping to, typically a few minutes each, and label them with what the listener learns, not with what was asked.

How chapters appear to listeners depends on where they listen. Some apps read chapters embedded in the audio file, some read chapter data supplied through the feed, and some pick up timestamps written in the show notes text; video podcasts on YouTube use timestamps in the description. Check your host's documentation for what it supports. If your host inserts ads dynamically, ask how it handles chapter times, since inserted ads can shift where a topic starts for each listener.

A worked example: chapters for a 52-minute interview

Pulling quotes that hold up

A quote in show notes, on social media or in an audiogram is a public attribution, so it has to be exact. Use the transcript to find candidates: short, self-contained lines with a clear opinion or a surprising detail. Then play each one from its timestamp and check the wording, because speech recognition can substitute a similar-sounding word that changes the meaning.

The transcript will not tell you who said what. In a two-person interview it is usually obvious from the content who is talking, but confirm attribution by listening before you put a name next to a line.

Links and resources mentioned on air

Search the plain transcript for phrases that signal a reference: "check out", "I read", "the book", "link", "website", "paper", the names of tools. Make a list with the timestamp for each, look up the correct link, and include it in the notes. Listeners often come to show notes looking for exactly one thing that was mentioned in passing, and the transcript is the only fast way to find all of them.

Writing the summary last

It is tempting to write the summary first from memory. Write it last instead, after the chapters, because the chapter list shows what the episode actually spent its time on, which is often different from what the episode was planned to be about.

Keep it specific. "We talk about running a business" tells a browsing listener nothing; "A baker explains how a single wholesale contract nearly closed her shop, and what she changed" gives them a reason to press play. If you use an AI assistant to draft a summary from the transcript, check every claim in it against the transcript, because assistants can blend or invent details.

Mistakes and limits to watch for

• Timestamps from a pre-edit file that no longer match the published episode.
• Misspelled guest names. Check spellings against the guest's own website or profile, not the transcript.
• Quotes taken from the transcript without listening, especially in crosstalk where recognition is weakest.
• Summaries that oversell, promising topics the episode barely touches.
• Ad reads and sponsor segments included as chapters, which most listeners do not want to jump to.
• Publishing the raw transcript as the show notes. It is a useful attachment, not a substitute.

Show notes with mydubly, in more than one language

mydubly produces the timestamped transcript, plain text, and SRT and VTT files you can use for video clips of the episode, from an audio or video file on your device. The spoken language is detected automatically. If you choose a target language, you also get the transcript translated, which gives you a head start on show notes for a multilingual podcast feed; have a fluent reader adapt them rather than publishing them as they are. The timestamped transcript format is the one to work from for chapters, and the podcast transcription page covers using transcripts across a whole show.

Next step

For your next episode, transcribe the final mix in audio to text, build the chapter list from the timestamps, then fill in guest details, links and quotes, and write the summary last. After two or three episodes it becomes a routine you can finish in well under an hour.