Ruff Cutter
Turn interview transcripts into a documentary rough cut.
Free. Your API keys. Your Mac. Your footage stays yours.
Download for Mac
Turn interview transcripts into a documentary rough cut.
Free. Your API keys. Your Mac. Your footage stays yours.
Download for MacYou shot the interviews. You know the story's in there somewhere.
But it's hour three of a four-hour transcript and you haven't touched the timeline yet. That part. Getting from raw footage to something you can actually cut. That's the part nobody talks about.
That's what this does.
Drop in your files. The app pulls the audio and transcribes it via AssemblyAI. Clean, speaker-labeled transcript in a few minutes.
Tone, audience, story beats, moments that have to be in there. The more specific you are, the better it cuts.
The AI proposes a structure. You read it, adjust it, tell it what's wrong. When it's close enough, export to Premiere as an FCP7 XML. Open a timeline that's already started.
Claude Sonnet, Gemini Pro, or GPT. Your keys, your cost, your call. Switch between them and compare.
Clip boundaries snap to actual word edges, not estimated segment times. Fewer fixes in Premiere. (Turns out that matters more than you'd think.)
The AI flags clips that didn't make the cut and suggests where they might fit. Nothing gets buried.
Don't like the cut? Tell it what's wrong. It tries again. You keep going until it's close enough to actually work with.
FCP7 XML for Premiere, EDL, paper edit as plain text, narration suggestions. Take what's useful, ignore the rest.
Star your best soundbites, group by theme, generate title slates. Works for anything with a talking head.
No account. No subscription. No data leaving your machine.
Project files live on your Mac. API keys go in macOS Keychain. The AI sees transcript text, never footage. Same as any transcription service you'd already use.
No Ruff Cutter cloud. No Ruff Cutter server. An app on your computer that calls APIs you signed up for directly.
That's the whole thing.
The app is free.
You pay your AI providers directly. A typical 3-hour project runs roughly $0.50–$2.00 depending on the model and how many passes you make. AssemblyAI transcription is about $0.37 per hour of audio.
You'll need keys from:
Put them in once. After that it just works.
macOS 10.13 or later. Apple Silicon. Internet connection for transcription and AI calls.
DMG installer. Code signed and notarized by Apple.