Speech to Text
Transcribe a short English recording with a model running in your browser.
Free browser tool
No Matrix computer needed
✓ No account or card
✓ Use the result right away
✓ Task-specific limits below
Workspace
Use speech to text
Runs in this browser tab
Mono or stereo, up to 25 MB and 2 minutes. Input formats depend on your browser's decoder. Files stay on this device.
Transcript
English-only Whisper Tiny runs in this browser. First use downloads model files (up to 150 MB) and stores them in browser cache. Speed and accuracy depend on your device. Your recording is never uploaded.
How to use it
Choose a short recording, download the local speech model on first use, then review and export the transcript.
What to check
Accepts up to two minutes and 25 MB. The first run downloads model files; recognition can mishear accents, names, or noisy audio.
Review the result before using it in a published page, application, or agent configuration.
Quick answers
Frequently asked questions
Does Speech to Text send my recording to a server?
No. The browser downloads a speech model on first use, then processes your recording locally. Model availability and performance depend on your device and browser.
How long a recording can I transcribe?
Choose an English recording up to two minutes and 25 MB. Review names, numbers, and timing because automatic transcription can make mistakes, especially with noise or overlapping voices.
Keep exploring
Related free tools
Need this work to keep going?
Matrix gives agents a private cloud computer for files, connected tools, and work that continues after you close your laptop. The browser tool above is free; a Matrix computer requires a paid plan.