Labs · Speech
Local, private speech to text.
Record up to 30 seconds and watch it become text. The model runs in this tab. Nothing is uploaded and nothing is stored.
idlelocal · private · on-device
Where it runs
Your browser
Network after load
0 requests
Audio uploaded
None
The model is downloaded once and cached. After that, recording and transcription happen entirely on your machine. The browser never sends the audio anywhere.