00:01 · WHAT YOU GET
A full transcription studio, running on your own hardware.
Offline & private
Every byte stays on your device. No cloud, no accounts, no telemetry — ever. The only network request Transcriptor ever makes is downloading a model you picked.
99+ languages
Whisper transcribes speech in over ninety-nine languages, out of the box.
Any file format
MP3, WAV, MP4, MOV, and more. Drop a four-hour video and walk away.
Timestamped export
TXT, SRT, VTT, JSON, or a clean article — with frame-exact subtitle sync.
Fix it in place
Click a line to play it, edit the text inline, or re-run one line on a bigger model.
Beautiful native UI
A first-class desktop app on Windows, macOS, and Linux with live progress.
∞ · NO LIMITS
- No file-length cap
- No monthly quota
- No cost
00:02 · HOW IT WORKS
Three steps, start to finish.
- 00:00 01
Drop your file
Drag any audio or video file onto the home screen. Video? Transcriptor extracts the audio automatically and cleans up after itself.
- 00:02 02
Pick model & language
Choose a Whisper model (we recommend large-v3-turbo) and a language, or let it auto-detect. Models download once, then live on your disk forever.
- 00:05 03
Review & export
Watch the transcript stream in live. Click a line to play it back, edit the text in place, or re-run just that line on a bigger model — then export to TXT, SRT, VTT, JSON, or a clean article. Timestamps stay frame-exact.
00:03 · IN ACTION
See it work.
Drop a file, watch the transcript stream in live, then work it: click a line to play it back, and re-run a single line on a bigger model when the words are worth arguing about. Nothing here is a mockup — it is the app.
00:04 · GET IT
Download Transcriptor. It's free.
One small installer per platform — Whisper models download on demand the first time you transcribe, so the app itself stays lightweight.
00:05 · INSTALL
Two minutes to first transcript.
Transcriptor isn’t code-signed yet, so both operating systems will warn you once on first launch. Here’s how to get past it.
macOS
macOS 11 or newer. Apple Silicon and Intel.- 01 Open the downloaded
.dmgand drag Transcriptor into your Applications folder. - 02 The first launch will fail with “Transcriptor is damaged and can’t be opened.” It isn’t damaged — macOS blocks apps that aren’t notarized by Apple. Clear the quarantine flag once, in Terminal:
xattr -cr /Applications/Transcriptor.app - 03 Double-click Transcriptor in Applications. It opens normally from now on.
Windows
Windows 10 or 11, 64-bit.- 01 Run the downloaded
.exeor.msiinstaller — either works. - 02 Windows SmartScreen may warn “Windows protected your PC.” Click More info, then Run anyway. The installer is unsigned, which is what triggers the warning.
- 03 Launch Transcriptor from the Start menu.
Nothing else to install — ffmpeg ships inside the app. Whisper models download on demand the first time you transcribe.
00:06 · QUESTIONS
Frequently asked.
How accurate is it?
Transcriptor runs the same Whisper models that power professional cloud transcription — accuracy comes from the model you pick (we recommend large-v3-turbo), and unlike a chatbot, nothing is ever truncated, paraphrased, or dropped.
Why not just ask ChatGPT/Claude/Gemini to transcribe?
Chat agents run a speech model in the cloud: your audio leaves your machine, uploads are capped at minutes of audio, timestamps are guessed, output can be silently truncated, and heavy use is metered. Transcriptor is unlimited, free, handles 4-hour videos, emits exact SRT/VTT-synced timestamps, and nothing ever leaves your computer.
What languages does it support?
Whisper transcribes over 99 languages and can auto-detect the spoken language. You can also lock a specific language for cleaner results on noisy audio.
Is it really fully offline?
Yes. Transcription runs entirely on your own hardware. The only time Transcriptor touches the network is when you choose to download a model — after that, you can go fully offline forever.
Which model should I choose?
Models range from tiny (fastest, least accurate) to large-v3. We recommend large-v3-turbo as the best balance of speed and accuracy; distil-large-v3 is the fastest English-only option. Each is offered in Compact, Balanced, or Full quality.
Is there any file-length limit?
None. A four-hour video transcribes end to end — nothing is capped, chunked away, or dropped. Longer files simply take longer.
Are the subtitle timestamps exact?
Yes. SRT and VTT timestamps align to the media timeline by construction, so captions stay in sync with your video frame for frame.
Do you collect any data?
No cloud, no accounts, no telemetry, no analytics. There is nothing to sign up for, and your audio, transcripts and library never leave your disk. The app makes exactly two network requests, both of which you control: downloading a model when you pick one, and asking GitHub whether a newer version exists on launch — a version check you can switch off in Settings. Neither sends anything about you or your files.
What does it cost?
Nothing. Transcriptor is free — no subscription, no per-minute metering, no paywalled features.