Transcribe many recordings at once
Each recording is turned into text with Whisper, the language found by itself. Pick SRT or VTT in the step for subtitles with timings. The AI model is downloaded once, the first time, and kept on this device; after that it starts at once. Drop the files below, check the preview, then run it on all of them. Nothing is uploaded.
Recipes
- Pictures to WebP
- Pictures to JPG
- Resize pictures
- Remove backgrounds
- Blur faces
- Upscale pictures
- Restore old photos
- Colourise photos
- Pictures and PDFs to text
- Transcribe recordings
Saved on this device
Preview
Files
Add a step
Or start from a recipe
Transcribed on this device, never uploadedManage models
How it works
- Drop your recordings. Drop files or a whole folder on the page. Each one is read by your browser and never uploaded.
- Check the preview. One of your files is shown before and after, so you can change the settings before running.
- Run and save. Run it on every file, then save them as one ZIP, into a folder, or one by one.
Questions
MP3, M4A, WAV, OGG, FLAC and the sound of MP4, MOV and WebM videos.
Yes: switch on Translate into English in the step.
Fast, the default, suits clear speech. Better makes fewer mistakes with accents and noise, and is slower and a bigger download.
No. Every file is worked on by your browser on your own device. Nothing is sent to a server, and there is no account.