Speech to text
Speech input is controlled by the keyboard’s microphone button. Enabling speech in Settings makes the feature available; it does not itself open the microphone.
Set it up
Section titled “Set it up”- Open Settings > Speech.
- Select the microphone you want to use.
- Choose Whisper or Windows speech.
- Choose a language option.
- If using Whisper, download the desired model.
- Press the microphone button to begin a listening session.
Whisper models
Section titled “Whisper models”| Model | Approximate download |
|---|---|
| Tiny | 78 MB |
| Base | 148 MB |
| Small | 488 MB |
Small is the current default. The interface reports whether the selected model is downloaded, downloading, or unavailable. Model downloads need an internet connection.
The application can fall back to basic Windows speech while Whisper is unavailable or downloading. Read the status message rather than assuming that the selected model is active.
Sending dictated text
Section titled “Sending dictated text”The speech settings offer an After I press Enter review mode. In that mode, dictated text is held for review instead of immediately extending the published chatbox message. Send or clear the message when the composer reaches its limit.
Microphone states
Section titled “Microphone states”- Off: not listening.
- Loading: Whisper is loading.
- Listening: the companion reports that capture is active.
- Thinking: the microphone has closed while the final words are processed.
- Attention: read the status text for the problem.
The speech workflow stops after 60 seconds of silence. If it says speech is not running, check Settings > Speech and the companion process before retrying.