PowerCut for Mac
Your assistant editor. Drop in raw talking‑head footage. PowerCut transcribes it on your Mac, strikes the silences, the ums and the repeated takes, and hands you a clean first cut you edit like a document.
One‑time purchase · 14‑day money back
Before any of the fun editing starts, someone has to scrub every minute of raw footage, find the good takes and cut the dead air. That someone used to be you.
PowerCut strikes the dead air, fumbles and bad takes first, so the footage you scrub is already mostly keeper.
Retakes are found and grouped, the best full read wins, with the reason shown and every alternate one click away.
Fillers are confirmed by two independent engines and pauses are cut on the audio itself, never clipping a word.
From a folder of raw clips to a timeline in the editor you finish in.
Step 1
Any video files, any length, multiple clips. PowerCut reads frame rates and timecode, then transcribes everything on your Mac.
A 20‑minute clip transcribes in about 20 seconds on Apple Silicon. MXF imports keep camera timecode and every audio track; exports reference your original files, untouched. MKV from OBS: File → Remux Recordings, then import.
Step 2
Silences, filler words, false starts, trail‑offs, production chatter and repeated takes get struck, with the best read of every line kept and every decision labeled with its reason. Nothing is deleted, everything is one click to restore.
Step 3
Edit like a document: strike a sentence, swap a take, punch in on a line, trim a clip edge to the frame. Then export a cuts‑only timeline to the editor you finish in.
Zooms are included as real transform adjustments, and captions match the cut and follow the voice: a new cue on a breath, a sentence end or a change of speaker, never a wall of text. Need to hand the edit to someone without the footage? Save Project with Audio writes one small folder that opens anywhere with the cut and audible playback.
No upload, no processing queue, no per‑minute meter. Your Mac does the work, and Apple Silicon does it absurdly fast.
The whole workflow, from raw footage to a timeline in your editor.
Everything runs on your Mac, every automatic decision shows its reasoning, and the app learns your taste the more you overrule it.
The transcript is the edit. Select words, strike them, and the timeline follows to the frame.
Try it: click any word to strike or restore it.
The transcript is the edit. Select words and strike them, press Return to split, drag sentences to reorder. Find anything with ⌘F, fix a mis-heard name everywhere in one click, and teach it corrections it applies to every future transcript. Press play and the text follows along.
Your footage never leaves your Mac. No uploads, no queue, no per‑minute meter, and it works offline.
Your footage never leaves your Mac. No uploads, no processing queue, no per-minute metering. Works offline, on a plane, and on client footage under NDA.
Repeated takes are grouped and the best full read wins, with the reason shown and every alternate one click away.
Repeated takes are grouped and the best full read wins, guided by your script if you load one: paste it, or just drop the Word file in. Every pick shows its reason, and every alternate read is one click away.
The automation tells you which calls it was least sure about, so you can settle each one in a click.
The automation tells you which calls it was least sure about: close-call take picks, context strikes, possible mishears. Audition each one in place and settle it in a click.
Your transcript as a clean page. Click any word to play, show or hide the cuts, and export it as a Word doc.
Audible skimming, trim brackets with live waveforms, ripple edits and frame‑precise overrides on any strike.
Audible skimming to find words by ear, FCP-style trim brackets with live waveform reveal, ripple edits, a range tool for surgical cuts, and frame-precise overrides on any strike. See the edit as a Rough Cut or every file as Full Footage, and let Follow keep the word that is playing in the middle of the strip at any zoom.
Every Auto Edit Pass saves a version first. Revert outright, or merge the differences one by one.
Every Auto Edit Pass saves a timestamped version first. Revert outright, or merge: review each difference between then and now, and keep the side you prefer.
Press F, K or M on any words. They show on the timeline and export as real Final Cut Pro metadata.
Press F, K, or M on any words, in the transcript or the Doc view: favorite the lines you love, keyword the b-roll to find later, and drop a ⚑named marker right on a word. They show on the timeline, jump from a searchable list, export as real FCP metadata on your original clips, and ship as a CSV log with timecodes and the line each one sits on.
Attach the other cameras and recorders, sync by timecode or waveform, and export synced multicams.
Attach the other cameras and recorders to each clip, sync by timecode or by waveform, and the export lands in Final Cut Pro or DaVinci Resolve as synced multicams, ready to angle-switch. Wrap every source in its own compound clip, or the whole shoot in one. Built the multicam in your NLE already? Import it and PowerCut links the angles, reads every dialogue angle as one conversation, and sends it back out as a multicam.
PowerCut labels who is talking, and can switch the picture to whoever is speaking.
Two hosts, an interview, a panel: tick Multiple speakers and PowerCut labels who is talking, from separate microphones, one track or angle per person, or from the voices on a single mic. Names sit in front of each turn in the transcript, in the Doc view, and in the captions if you want them. Give each speaker a camera and the picture switches to whoever is talking; the multicam export carries every switch.
Every cleanup is computed in the background, so flipping one applies or lifts it instantly.
Every cleanup pass computes in the background whether or not its box is checked, so flipping one applies or lifts it instantly. Spots you edited yourself are always skipped: your edits win. Music stays in: PowerCut listens to the channels and tracks the transcript does not use, keeps the pauses your music fills, and a Music Tutorial style is built for voiceover over playing.
Select the words that deserve a closer look and zoom in. Zooms export as real transform adjustments.
Select the words that deserve a closer look, or right-click a clip on the timeline, and zoom in: two editable preset levels, a draggable dot to aim it and a ring to set the strength, live in the preview. Zooms export to Final Cut Pro and DaVinci Resolve as real transform adjustments, frame accurate at any resolution.
It learns you, privately. Every strike you restore teaches PowerCut. Rules you consistently overrule get more conservative on your Mac, for your footage. Nothing is collected, nothing is sent anywhere.
PowerCut transcribes and cuts in twenty languages, entirely on your Mac. It sets the language from your first clip, and any clip can carry its own.
These six run the full pipeline: transcription, pauses, filler words, stutters and repeated takes. Models download once, on first use, and nothing is ever uploaded.
Eight already run the full pipeline, filler words included. The others get transcription, pause cuts and the whole paper edit.
Japanese, Russian, Ukrainian, Polish, Dutch, Swedish, Lithuanian and Greek run the full pipeline too, filler words included; Korean, Chinese, Cantonese, Hindi and Hinglish get transcription, pause cuts and the whole paper edit, with the filler-word pass still to come. Hindi, Hinglish and Vietnamese are carried by an on-device Whisper engine (a one-time model download on first use). Cantonese and Chinese run on Apple's engine or, with the Maximum or Large v2 accuracy choice, on Whisper, with a second read on another model when the first comes back unusable; Cantonese writes Traditional or Simplified, your pick. Need another language? Email apps@rafaelludwig.com, requests decide what ships next.
Shooting in a language you don't edit in? PowerCut can transcribe the speech and write the transcript in English, with every word timed to the original footage, so your cuts land exactly where things were said.
Translate one clip from its language menu or the whole project with the Translate button; name the spoken language or let Auto Detect handle mixed footage. Translating an already-edited clip keeps your cut in place. On-device like everything else, and a Maximum accuracy option runs OpenAI's full Large-v3 model when the wording matters.
The full editor, forever. Export locks after the trial.
One-time purchase with 18 months of updates.
One-time purchase. Every update, forever.
Requires an Apple Silicon Mac with macOS 26 (Tahoe) for on‑device transcription. On macOS 14 and 15 the app opens and edits existing projects, but cannot transcribe new footage.
The same rough‑cut workflow on Windows, with frame‑accurate export to DaVinci Resolve and Premiere Pro. Join the waitlist and you will be first into the beta, with early‑bird launch pricing locked in.
Updates & ownership. Both licenses are permanent and never expire. PowerCut includes every update released in the 18 months after purchase, then keeps working on the last version you received; upgrading to Lifetime later is always available. Lifetime includes every update for as long as PowerCut exists. If a product is ever retired I will say so rather than go quiet: the final version stays downloadable and your license keeps working. Refunds: if it is not for you, email apps@rafaelludwig.com within 14 days of purchase and you get your money back, no questions asked.
No. Transcription, take detection, the context pass, everything runs on your Mac using on-device models. PowerCut works with the network cable unplugged. The only network calls are license activation and the update check. The one exception is opt-in: if you connect the Claude Desktop app or Codex to PowerCut (Assistant Editor +, beta), the transcript (never audio or video) goes to that AI through your own app and account, and only when the AI asks for it.
A cuts-only, frame-accurate timeline, landscape or vertical to match the footage. Final Cut Pro and Resolve get FCPXML, as plain clips or wrapped in compound clips or multicams, one per source or one for the whole shoot; re-exports merge into the same FCP event. Premiere gets an XML where your full footage lives in a nested sequence and the edits sit on top, the way Premiere editors expect. Markers, keywords, and favorites travel too: FCP browser metadata on the original clips, marker EDLs for Resolve, timeline markers for Premiere. Captions export as .srt timed to the edited cut.
Never. Your files are referenced in place and untouched. The exported timeline points at your original media, and every automatic cut inside PowerCut is a restorable strike, not a deletion.
An Apple Silicon Mac. On-device transcription uses Apple's speech engine on macOS 26 (Tahoe). The verbatim engine downloads once (about 500 MB) for filler-word and stutter detection.
English, Spanish, French, German, Italian and Portuguese with the full pipeline, plus Japanese, Korean, Chinese (Simplified and Traditional), Cantonese, Hindi and Hinglish in beta. Japanese gets the full pipeline too; the other beta languages get transcription, pause cuts and the whole paper edit, with filler detection still to come. The six main languages are fully supported; the rest are labeled beta while real-world feedback shapes them. Pick the project's main language on the start page, or set a language per clip for bilingual shoots. Language models download once, on first use, and everything still runs on your Mac. And any language can be translated to English while you edit (see the next question), while Auto Detect handles clips that switch language mid-recording, writing each part in its own script. More languages are coming: email apps@rafaelludwig.com and tell me which one, requests decide what ships first.
Yes, in beta. Choose Translate to English from a clip's language menu (or the Translate button for the whole project), and Whisper transcribes the spoken language and writes the transcript in English, on your Mac. Every English word stays timed to the original footage, so striking a sentence cuts exactly where it was spoken, and if you translate a clip you already edited, your strikes and cuts stay in place. Pick the spoken language for best accuracy or use Auto Detect; a Maximum accuracy option runs OpenAI's full Large-v3 model (a one-time download of about 3 GB). Translation always writes English; that is the one language Whisper translates into.
An optional layer of AI help that uses the AI app you already have. PowerCut works fully without it. Connect the Claude Desktop app or Codex in Settings (no key, no sign-up), then chat with it about the project you have open: it reads the transcript, strikes and restores, corrects misheard words, picks takes, settles review items, adds markers and runs the Auto Edit Pass, and you watch every change land in PowerCut with the assistant's reason and an Undo. Only the transcript and PowerCut's own measurements ever go to the AI, through your own app and account (never audio, video or files), and only when the AI asks for them. It runs on your Claude or ChatGPT plan; PowerCut charges nothing for it. ChatGPT's desktop app cannot connect yet (it only takes remote connectors).
Seven days fully functional, export included. After that the editor keeps working forever, and a one-time license unlocks export whenever you're ready. Your work is never held hostage.
Yes. If PowerCut isn't for you, email apps@rafaelludwig.com within 14 days of purchase and you get your money back, no questions asked. With the free trial before buying, that's three weeks to be completely sure. Refunds go back to your original payment method, and the license key is deactivated.
One‑time purchase · no subscription · 14‑day money back
Buy Lifetime • $149.95 Buy PowerCut • $79 Free Download Windows • Join WaitlistWindows • Get Notified!Already own PowerCut? My Purchases
Share PowerCut and the other PowerApps with your audience and earn a commission on every sale through your link. Free to join, 30-day tracking, monthly PayPal payouts.