Dictation
Dictate messages in Nested with the browser’s speech service or Whisper running on your own device, and tune language, model, and your custom words.
Select Dictate in the composer to speak instead of type. Settings → Dictation controls how that works on this device, including whether your voice ever leaves it.
Before you start#
- Open Settings → Dictation.
Steps#
- Choose the Speech engine: Automatic (Nested picks one), Browser speech service (in Chrome and Edge, this sends your voice to Google or Microsoft to turn it into text), or Whisper on this device (your voice never leaves the device; the model downloads once).
- Set the Language you speak: Same as the app, English, or Persian. It can differ from the app language.
- For Whisper, pick a Whisper model: Tiny (about 40 MB), Base (about 75 MB, recommended), or Small (about 250 MB). Larger models understand more and use more memory and data.
- Choose Run Whisper on: Graphics card (faster) or Processor (works everywhere).
- Add names and terms under Your words so Whisper spells them your way. Words you fix after dictating are added automatically.
- Under Downloaded files, select Delete downloaded files to free space; Whisper downloads its model again the next time you use it.
Diagnostics shows how dictation behaves on this device — counts and timings only; no audio or text is kept.
Example#
Choose Whisper on this device with the Base model, add your product and teammates’ names to Your words, and dictate task updates without your voice leaving the laptop.
Questions
- Does dictation send my voice to the cloud?
- Only with Browser speech service in browsers like Chrome and Edge. Whisper on this device keeps your voice on the device.