What it does
Offline Transcribe turns audio and video files into text with OpenAI's Whisper speech models, run on your PC through whisper.cpp. It uses your graphics card when it can and your processor when it cannot.
- Common audio and video files, such as MP3, WAV, M4A, FLAC, MP4 and MOV. Some formats, such as OGG or MKV, need Windows' codec extensions or ffmpeg installed on the PC.
- 29 languages, chosen by you or detected automatically.
- Paragraphs with timestamps. Copy the text, or save it as TXT, as SRT or VTT subtitles, or as a Word file (DOCX). Word does not need to be installed.
- Choice of model. Smaller models are faster; larger ones are slower and usually more accurate. Each model is downloaded once, the first time you use it, and checked before use.
Free and Pro
| Feature | Free | Pro |
|---|---|---|
| Length per file | First 10 minutes | Whole file |
| Files at a time | One | A queue of files, or whole folders |
| Models | Tiny, Base and Small | Also Medium, Large v3 Turbo and Large v3 |
| Copy and TXT | Yes | Yes |
| SRT and VTT subtitles | No | Yes |
| Word (DOCX) | No | Yes |
Your data
Your recordings and transcripts stay on your PC; they are never uploaded. The app goes online for three things only: to download a speech model from Hugging Face the first time you use it, to check a Pro key with blznworks.com, and to send usage counts that carry just the product, the channel and the event. The privacy policy has the details.
Availability
Offline Transcribe needs Windows 10 version 2004 or later on a 64-bit (x64) PC. Get Offline Transcribe from the Microsoft Store. The free edition needs no key; to unlock Pro, enter your key in the app. For questions, use the contact form.
Price
The free edition needs no key. Offline Transcribe Pro costs:
- $29 once, with no renewal
One Pro key works on up to 3 devices. Paddle handles the payment as merchant of record and shows any tax before you pay.