Podcasts
Subtitle podcast clips without re-uploading the episode.
Ninety-minute episodes, a transcript you fix once, and burned-in captions on every clip you cut out of it.
To subtitle podcast clips, caption the master episode once and cut from the corrected transcript — that way guest names, company names and jargon are fixed a single time instead of once per clip. LumaCaption takes episodes up to 90 minutes and never uploads the video file.
A podcast produces one long recording and a dozen short clips, and the captioning cost lives almost entirely in the clips. Most workflows pay that cost twelve times: upload each clip separately, wait for each transcription, and correct the guest’s company name in every single one.
The order that saves the work is the other way round. Caption the episode once, correct the names and the jargon once, and cut clips out of a transcript that is already right. What you fix in one pass stays fixed everywhere it appears.
Whole episodes, not chunks
Pro handles 90 minutes per video and Creator 30 — most interview episodes fit in one file, so timings never have to be stitched back together.
Fix a name once
Guest names, company names and technical terms are edited in the transcript, and every clip cut afterwards inherits the correction.
Exports never cost minutes
Transcription is what is metered; rendering is not. Cutting twelve clips out of one captioned episode costs nothing beyond that first pass.
A style per clip
240+ styles, and any single line can wear its own look — so a bold hook and a clean body can live in the same thirty-second cut.
The minute arithmetic, honestly
Transcription is charged in minutes of audio, and there are two engines: Candy Lite, which is generous, and Candy Ultra, which is the accurate one. A 60-minute episode is 60 minutes of transcription whichever engine you use — so the question worth asking before you start is which engine an interview actually needs.
Ultra is worth it when the audio is difficult: two people talking over each other, a remote guest on a bad line, heavy accents, or Hindi and English mixed together in the same sentence. Lite is fine for a clean, close-miked solo recording. Pro includes 20 hours of Ultra a month, which covers roughly twenty hour-long episodes; Creator includes 3 hours.
Exports are the part that is free. Rendering a clip does not draw down minutes at all, on any plan — so the twelve clips you cut from a captioned episode cost exactly nothing after the episode itself.
- Candy Lite
- The free-tier engine. Fine for clean solo audio. 20 minutes a month on Free, 10 hours on Creator, 50 on Pro.
- Candy Ultra
- The accurate engine, for crosstalk, remote guests, accents and code-switching. 5 minutes on Free, 3 hours on Creator, 20 on Pro.
- Exports
- Free and unlimited on every plan. Re-rendering after a correction never costs minutes.
- Top-ups
- Prepaid Ultra minutes if a month runs long, valid twelve months and drawn only after the monthly allowance is spent.
What makes a podcast clip get watched
A podcast clip has no visual event in it. Two people sit and talk, and the frame does not change for thirty seconds. That is why captions are not optional here in the way they are optional on a cooking video: the text is the only thing on screen that moves.
The consequence is that caption style does more work in a podcast clip than anywhere else. Word-by-word reveals give a static two-shot the motion it does not have. A full-frame type scene can take over entirely for a punchline, then hand back to the footage. Both are one click, and both can be applied to a single line rather than the whole clip.
How it works
Add the episode
Up to 90 minutes on Pro. The recording is never uploaded — only its audio is used to find the words.
Correct names and jargon once
Guest names, companies, products, acronyms. Every clip you cut afterwards inherits the fixes.
Cut, style and export each clip
Trim to the moment, pick a style per clip, export burned-in. Rendering is free on every plan, however many clips you cut.
Common questions
Can I caption a full-length podcast episode?
Yes — up to 30 minutes per video on Creator and 90 minutes on Pro. Captioning the master and cutting clips from the corrected transcript is the workflow that saves the most time, because guest names and jargon get fixed once instead of once per clip.
Does exporting twelve clips cost twelve times as much?
No. Transcription is what is metered, and exports are free and unlimited on every plan including free. Once an episode is transcribed, every clip you cut and render from it costs nothing.
Which engine should I use for a two-person interview?
Candy Ultra, if the audio has crosstalk, a remote guest, strong accents or Hindi-English code-switching — that is exactly where the accuracy gap shows. Candy Lite is fine for a clean, close-miked solo recording. Pro includes 20 hours of Ultra a month, Creator 3 hours.
Does it separate speakers?
The transcript is a single word-timed stream rather than a diarised, per-speaker one. For clip captions that is what you want — the caption follows whoever is talking. If you need a labelled who-said-what transcript for show notes, this is not the tool for that part of the job.
Is my recording uploaded?
No. The video file never leaves your device; only the audio is used to work out the words and their timings. For unreleased episodes and embargoed guests, that is usually the deciding factor.
