Whisper in your browser · SRT / VTT / TXT · zero uploads
Transcription that never uploads your voice
Drop an audio or video file and get timestamped, downloadable transcripts — the Whisper speech model runs inside your own browser via WebAssembly. Free, private, completely local.
One tool, zero sign-up
Built for the moment a recording has to become text: lectures, interviews, voice memos, podcast drafts — and captions for the video you are about to publish.
FAQ
Is the transcription really free?
Yes — files up to five minutes long, with SRT and plain-text export, are free with no signup. TranscribeKit Pro ($3.99 once) removes the length cap, adds WebVTT export and paragraph mode.
Is my audio uploaded anywhere?
Never. The Whisper model runs inside your browser via WebAssembly after a one-time download from this same site. There is no upload endpoint — your recordings cannot leave your device, and the tool keeps working offline once the model is cached.
How accurate is it?
It runs the Whisper tiny English model — fast, and accurate for clear speech in a quiet room. Heavy accents, noisy rooms or music beds mean more errors; the timestamped segments make manual fixes quick, especially once you pass the file to SRT Studio for editing.
Which formats can I transcribe?
Anything your browser decodes as audio: MP3, WAV, M4A, OGG, FLAC, WebM and the audio track of MP4 files. Output is timestamped segments with SRT, VTT and TXT export.
What exactly does Pro unlock?
No length cap (free handles up to 5 minutes per file), WebVTT export and paragraph-mode transcripts. One payment of $3.99, unlocked on this browser forever.