Free audio and video transcription
Create speaker-labeled transcripts from audio or MP4 video in 25 European languages. Your recording never leaves this page. Speech models download from Weeve’s CDN.
Drag audio files here
or
From recording to transcript
Five steps run in your browser, from local decoding to export.
- 01
Decode locally
Your browser reads the chosen audio or video file and resamples a mono copy to 16 kHz.
- 02
Download and cache models
Speech and speaker models come from Weeve’s CDN and are cached by the browser when storage allows.
- 03
Transcribe speech
NVIDIA Parakeet creates punctuation, capitalization, and word timestamps in one of 25 languages.
- 04
Identify speakers
sherpa-onnx groups speech into speaker labels. You can correct the count from 1 to 8.
- 05
Review and export
Rename speaker labels, copy the transcript, or export Markdown, plain text, or PDF.
Product details
The limits and requirements to know before you start.
- Accepted files
- WAV, MP3, M4A, MP4, OGG, OGA, FLAC, and WebM. The browser must support the codec inside the file.
- Recording limit
- Each recording can contain up to 60 minutes of audio.
- Browser support
- Use a recent Chrome, Edge, or Safari with WebAssembly SIMD and Web Audio support.
- Local acceleration
- WebGPU is used when available. WebAssembly is the fallback.
- First model download
- About 1.3 GB for WebGPU or 0.94 GB for WASM. Cached files can be reused on later visits.
- Speaker fallback
- If speaker separation fails, the completed word transcript remains available without speaker labels.
25 languages, detected automatically
The model adds punctuation, capitalization, and word timestamps without a language picker.
- Bulgarian
- Croatian
- Czech
- Danish
- Dutch
- English
- Estonian
- Finnish
- French
- German
- Greek
- Hungarian
- Italian
- Latvian
- Lithuanian
- Maltese
- Polish
- Portuguese
- Romanian
- Russian
- Slovak
- Slovenian
- Spanish
- Swedish
- Ukrainian
Your recording stays here
Recording bytes, decoded audio, transcript text, and speaker names stay on this page. Speech models download from Weeve’s CDN.
A transcript you can review
Timestamps and speaker labels stay editable before export.
Maya
Let’s confirm the launch checklist before we finish.
Speaker 02
The help article is ready. I will send the final link this afternoon.
Maya
Great. I renamed your speaker label, so the exported notes will use your name.
Frequently asked questions
Practical answers about files, processing, privacy, and results.
