Scribello
Your recordings, written out — privately, on your own computer.
Drop an interview, a lecture or a meeting. Minutes later you have a clean transcript — who said what, every name spelled your way, every word one click from its sound.
As simple as it should be
Drop a recording
Any audio or video file: MP3, M4A, WAV, MP4, MOV and more.
Let it listen
Follow the text appearing live. Long recordings? Keep working, it runs in the background.
Read, polish, export
Fix a word as in any document, then export to Word, plain text or subtitles.
Everything a transcript needs
The care of a professional tool, the simplicity of a notebook.
Top-quality recognition
Whisper large-v3 runs on your machine, using the graphics chip when there is one. Dozens of languages, detected automatically.
Speakers told apart
Paragraphs are labelled by voice. Name someone once and the name follows them everywhere.
A glossary of your words
Names, jargon, foreign terms: spelled exactly the way you write them, diacritics included.
Click a word, hear it
Every word is tied to its moment in the audio. The word being spoken lights up as you listen.
Shortcuts that work in Word
Play, pause and step back from any app, so you can type up while you listen.
Find anything
Search every transcript at once. Accents and capitals don't matter.
Names, spelled your way
Speech recognition stumbles on names and specialist words. Scribello doesn't have to.
Add the terms of your field to the glossary — or import a list. Scribello listens out for them, then corrects near-misses to your exact spelling. Every correction is underlined, and one click puts back what it heard.
Nothing leaves your computer
Interviews, therapy notes, family memories, confidential meetings: your recordings are processed entirely on your own machine. No account, no cloud, no subscription.
Scribello only goes online to download its speech model once, from Hugging Face, and to check GitHub for new versions — which you can turn off. No analytics, no telemetry.
Download Scribello
Free, for good. The speech model (1.6 to 3.1 GB) downloads once, on first launch.
All versions on the releases page.
Windows says it “protected your PC”?
- Click More info.
- Click Run anyway.
Windows shows this for new apps that have not yet built a reputation with Microsoft. Scribello is open source and built in public on GitHub; the Mac version is signed and notarized by Apple.
Linux: the AppImage does not start?
Make it executable once — right-click → Properties → Permissions → “Allow executing file as program”, or in a terminal: chmod +x Scribello-Linux.AppImage.
Free, and made with care
Scribello is free. If it earns its place in your work, you can support my work. I hope that the tools I build will help the Dharma flourish — translation, publication, the practical work that surrounds them. I also back a few other causes and organizations I believe in.