Meet Nemotron, a new on-device speech engine for Windows and Android. It transcribes in real time, entirely on your device.
  • Nemotron Multilingual transcribes around 40 languages locally
  • Offline speaker separation (diarization) so you can see who said what
  • Import an existing audio file and transcribe it
How to turn it on:
  1. Open Settings -> Speech & AI
  2. Under Speech Engine, choose Nemotron Multilingual (or Nemotron English Only)
  3. Tap Download Nemotron model (~600-700 MB) and wait for 'Nemotron model ready'
Nemotron is in Beta. If a language comes out wrong, switch back to Whisper in the same screen.