Transcription & the Text-Based Edit
Transcription is the gateway AI skill — the one that makes every other assist in this course possible. Learn it and your footage becomes a searchable document you can cut by deleting words. Here's where that superpower shines, and exactly where it falls on its face.
If you learn exactly one thing from this course, learn this chapter. Transcription is the gateway AI skill of post-production — the unglamorous one that everything else stands on. The moment your footage is transcribed, every clip becomes searchable by what was said in it. Think about what that means on a real project: you don't scrub for the moment your subject mentioned her father, you type "father" and you're there. Your media browser becomes a search engine for your own movie.
And transcription unlocks the strangest, most useful new craft in modern editing: text-based editing. Both DaVinci Resolve and Premiere Pro now do this natively — the transcript appears next to your timeline, and when you delete a sentence in the text, the corresponding video disappears from the cut. You are editing footage the way you edit a document. The first time you watch a paragraph deletion ripple through a timeline, something in your brain permanently rearranges.
The day-one workflow
Here's the habit that makes it all work — and the discipline is entirely front-loaded:
- Transcribe everything on ingest, day one. Not when you need it — before you need it. It's a batch job; run it overnight while the cards are backing up.
- Skim and correct the transcripts. Fix names, jargon, and mangled phrases now, while the shoot is fresh — a wrong word in the transcript is a moment you'll never find again by search.
- Read before you watch. Reading interviews is many times faster than watching them; highlight the transcript like a script and you've built a paper edit before touching a timeline.
- Cut the radio edit in text. Assemble the story by keeping and deleting sentences — structure first, judged by ear, story logic only.
- Then switch to the timeline. Refine every cut point by eye and ear: reactions, breaths, b-roll, rhythm. The text got you to a draft; the timeline makes it a film.
That last step is not optional, and it's where the honest part of this chapter lives.
Where it shines, where it fails
Text-based editing shines wherever the content is the words: documentaries, interviews, dialogue-driven scenes, corporate and educational work, podcasts with cameras. For a doc editor it's close to a superpower — the paper edit, that decades-old documentary tradition of structuring a film on the page, now snaps directly into a working timeline instead of living in a binder. Dialogue assemblies that took a week happen in an afternoon.
And it fails, hard, wherever the content isn't words. Action. Visual comedy. A reaction shot — the cutaway to the person listening, which is frequently the most important shot in the scene and appears nowhere in any transcript. Cut a conversation purely in text and you'll get a competent radio edit wearing pictures: every line present, every look missing. The transcript also can't hear performance — it renders the flat read and the devastating read as identical sentences. So use text-based editing for what it is: the fastest structural tool ever handed to an editor, and a completely untrustworthy finishing tool. Structure in text, finish on the timeline, always.
First doc I cut with day-one transcription, the director called mid-edit: "Didn't somebody mention the flood at the old house?" Old workflow, that's an afternoon of scrubbing three interviews on a maybe. I typed "flood," found it in eight seconds — buried in an answer to an unrelated question, in an interview neither of us remembered it from. It became the opening line of the film. That moment exists because the search existed.
Once everything is transcribed and searchable, the next move is obvious: let the machine take the first pass at the whole mountain. Next chapter — AI selects and the rough cut you treat as a draft, never a decision.
Recut is this course in tool form — an AI edit assistant that pulls selects from your footage, builds a first assembly, and preps cutdowns, ready to polish in your own editor.