How to Transcribe a Podcast
Why every episode should have a transcript, how to produce one efficiently, and how to turn it into show notes, captions and a searchable archive.
The Scriptorium team ·
A podcast transcript does three jobs at once: it makes the episode accessible, it gives search engines something to read, and it is the raw material for show notes. One pass, three outputs.
Why bother
- Accessibility. Deaf and hard-of-hearing listeners cannot access your show without one. That is the whole reason, and it is sufficient on its own.
- Discoverability. Audio is invisible to search engines. A transcript on the episode page is the only text they have to work with.
- Show notes. Far faster to edit down from a transcript than to write from memory.
- Your own archive. After thirty episodes, finding the one where a guest said a particular thing is impossible without text.
- Listeners who prefer reading. A meaningful share of people would rather skim than listen, and currently they simply do not.
The workflow
- Transcribe the finished episode. Upload the published audio to audio to text — the edited version, so the transcript matches what listeners hear.
- Label the speakers. Rename host and guests once.
- Correct it. Read along with the audio. Prioritise guest names, company names, book titles, products and any technical vocabulary.
- Export twice. Markdown for the episode page, WebVTT if you also publish video.
Do this as part of publishing an episode rather than as a separate project. A backlog of untranscribed episodes never gets done; one episode at publication time takes twenty minutes.
Getting a cleaner starting point
Podcast audio is usually good, which is why podcast transcripts are usually good. The exception is remote interviews. If you record guests over a call, record each side locally — a double-ender — rather than relying on the call recording. It eliminates crosstalk between tracks and removes connection artefacts, which improves the transcript as much as it improves the audio.
Publishing it
On the episode page
Put the transcript on the same page as the player, below the show notes. Splitting it onto a separate page divides the page's relevance between two URLs and makes the listener click.
Format it
A wall of text nobody reads is a wasted transcript. Speaker names in bold, a blank line between turns, and headings at the topic changes. The Markdown export gives you most of this already.
Do not just paste raw
Spoken language on the page reads badly. Removing filler words and tidying false starts makes it readable without changing what anyone said.
Turning it into show notes
The transcript gives you, directly: the topics discussed with timestamps, every book, product and person mentioned, and the quotable moments. Show notes are mostly an editing job on top of that rather than a writing job.
Timestamped text is the useful export here, because chapter markers come straight out of it.
Video podcasts
If you publish to YouTube as well, export WebVTT and upload it as the subtitle track. Same transcript, no extra work, and better than the automatic captions YouTube would otherwise use — particularly for guest names.