Transcripts for editing
How to transcribe interview footage for editing.
Generate, correct, export the SRT, cut on paper.
Upload a light export of each interview, generate a caption track, read it once against picture to fix names and jargon, then export it as an .srt. That file is a timed transcript: you can search it, print it and build a paper edit from it before you touch the timeline. If you cut in Final Cut Pro on Apple silicon and your interviews are in U.S. English, the editor's own tool may be all you need.
Video hosting for filmmakers · 5 GB free · Paid plans from USD 9/month
By uncompressed.io · Updated September 2026
Check your editor first
If your editor already transcribes,
you may not need anything else.
Two editors document speech transcription of their own. Before uploading anything, check whether yours covers your footage.
Final Cut Pro
Apple says Final Cut Pro can detect speech and create subtitles automatically. The feature is listed as U.S. English only, requires a Mac with Apple silicon, and downloads a language model the first time, which needs an internet connection. If that describes your setup and your interviews, start there.
DaVinci Resolve
Blackmagic says the Edit page can transcribe clips for subtitles or text based editing, and imports timed text including SRT.
Where a hosted transcript earns its place is outside the editing seat: interviews in a setup those tools do not cover, a transcript you want to hand to a director or producer who does not run your editor, or an SRT file you want to read, print and mark up away from the timeline. The rest of this page is that workflow.
The workflow
From interview to paper edit,
four steps, one of them slow.
Three of these steps take minutes. The correction pass takes as long as it takes, and it is the one that decides whether the transcript can be trusted.
Upload each interview and generate
One film per interview: a browser-playable proxy (for example H.264) with the full audio and a burned-in source timecode window, so any line you pick can be traced back to the camera clip. Then click Generate captions. A speech-to-text service writes the track, only when someone clicks, and the runtime comes out of your plan's monthly caption minutes.
Correct against picture
The owner editor shows a transcript that follows playback. Click any line or timestamp to jump there, fix the text inline and save. Fix names, places, product names, figures and jargon first: those are what you will search for later.
Export the SRT
Download the corrected track as .srt, or .vtt if a web tool wants it. The file is plain text: numbered cues, each with a start time, an end time and the words.
Build the paper edit
Open the SRTs in any text editor or print them. Highlight the lines you want, add who is speaking, order them into a story, and note each line's timecode. Then assemble the timeline from that document.
For how the selects, the paper edit and the radio edit fit together, read what a paper edit is. For the correction pass itself, the caption accuracy checklist gives the order to work in.
Why the correction pass is not optional
Caption minutes
Interviews are long;
plan the shoot against the allowance.
Generation is metered per calendar month by plan. A documentary shoot can produce several hours of interviews in a week, so compare the runtime with the allowance before you start.
Capacity is arithmetic on a 45-minute interview, not a measured figure; your interviews will run longer or shorter. Minutes reset each calendar month. Plans also differ by storage and seats, and the free plan holds 5 GB, one film at a time. Yearly billing is available on every paid plan.
Finding the line later
Once it is corrected,
the transcript becomes an index.
A corrected track is useful long after the paper edit. In the review room, anyone with the share link can read the transcript, search the dialogue by word and jump to the moment it was said, with no account. That is how a director or producer finds “the line about the warehouse” without asking the editor for a timecode. The search matches the words in the corrected track, so a name the transcript still gets wrong will not be found.
Reviewers cannot overwrite your text. They submit a correction and you apply it, which is useful when the interviewee’s own team knows how a name or a product is spelled. The option is on by default and can be switched off per film.
For what is inside an SRT, and the two formatting faults that break hand-made files, read how to export an SRT file from a video.
When this is the wrong tool
Know the limits,
and say so.
Your editor already covers it
If Final Cut Pro's automatic subtitles or Resolve's transcription fit your footage, keep the transcript inside the editor you cut in. Uploading adds a step you do not need.
Timings are file-relative
SRT cue times count from the start of the uploaded export, not from camera timecode. The burned-in timecode window is what connects the paper edit back to the original clips.
The text is a draft until you read it
No transcript here comes with an accuracy promise. Overlapping speech, names and figures need a person reading against picture, every time.
Questions
Frequently asked
What file should I upload for transcription?
A light, browser-playable export of the interview (for example H.264) with the full interview audio, not the camera original. You will play it while you correct the text, so it should start instantly. It counts toward your storage; the free plan holds 5 GB, one film at a time.
Will the SRT timings match my camera timecode?
No. Cue times in an SRT count from the start of the file they were made against, so a cue at 00:12:04 means twelve minutes into the uploaded export, not source timecode. Burn a source timecode window into the export before uploading, and every line you pick in the paper edit can be found in the original clip by reading the window.
Can the director or producer search the interview without an account?
Yes. The review room opens from the share link with no account, and reviewers can read the transcript, search the dialogue by word and jump to the line. A reviewer who spots a misspelled name submits a correction and you apply it; reviewers do not overwrite captions. You can switch reviewer caption access off per film.
What happens when an interview is longer than my monthly minutes?
Minutes are counted per calendar month, and generation is refused once the month's allowance is used. Check the interview runtime against the allowance before you start. A reviewer can also ask for captions to be generated from the room, and those minutes come out of your allowance.
Will it transcribe my language well?
This page makes no promise about any language. Test before you commit a shoot: caption a short excerpt on the free plan (20 minutes a month) and read it against picture. The correction pass tells you more than any claim would.
Can I transcribe an interview I moved to the archive?
Not until it is restored. Generation is refused for an archived film, so restore it to hot storage first, or keep a light export of the interview hot while the edit is running.
Transcribe one interview, then decide
Caption generation is metered per calendar month by plan. The free plan includes 20 minutes and holds 5 GB, one film at a time, which is enough to test the workflow on one interview excerpt. Lite includes 10 hours a month.
Sources
- 1.Apple Support, Final Cut Pro User Guide: Generate subtitles automatically (fetched 17 September 2026; U.S. English only, requires a Mac with Apple silicon, language model downloaded on first use)
- 2.Blackmagic Design, DaVinci Resolve Edit page (fetched 17 September 2026; transcribe clips for subtitles or text based editing, imports TTML, SRT, XML and embedded MXF/IMF subtitles)
- 3.W3C Web Accessibility Initiative: Captions/Subtitles (fetched 17 September 2026; automatic captions must be edited for accuracy and can serve as a starting point)