Podcast transcript format: what good looks like
There is no single official standard — but there is a format that consistently works for readers, search engines, and screen readers. This is it, with before/after examples.
The five rules
- Speaker labels on their own line —
HOST:/GUEST:or names. Never inline mid-paragraph. - Timestamps at topic changes, not every line —
[12:45]every 2-4 minutes is scannable; per-sentence stamps are noise (keep those in your .srt instead). - Paragraphs of 2-4 sentences — machine transcripts arrive as a wall of text; break at breath points and topic shifts.
- Clean verbatim, not strict verbatim — cut fillers (um, you know) and false starts, keep every meaningful word. Strict verbatim is for legal transcripts, not show pages.
- Punctuate for reading, not speaking — spoken run-on sentences get split; em-dash pileups become periods.
Before / after
RAW: so um i think the the biggest mistake is you know people try to grow on every platform at once and like they burn out right and well what we did was we just picked one
CLEAN: I think the biggest mistake is people try to grow on every platform at once, and they burn out. What we did was we just picked one.
Format by destination
- Episode page / blog: clean verbatim + labels + sparse timestamps (this is also the SEO format — the page becomes indexable long-form content).
- Captions: keep per-chunk timestamps, export .srt or .vtt — don't reformat, players need the timing.
- Show notes: don't paste the transcript — summarize it: 2-3 sentence overview, takeaways with timestamps, links mentioned.
- Quotes for social: pull verbatim lines (cleaned), 1-2 sentences, with the timestamp so you can cut the clip.
Do this automatically
The free cleanup tool applies rules 3-5 for you — paste any raw transcript (or transcribe audio in your browser) and get clean verbatim text, .srt/.vtt captions, and show notes. Every removal is shown, nothing is stored.