Help · dictation
Voice training
You train the keyboard on lines you say. Sober speech and slurred speech keep separate corrections.
Where it is
- On the Tank, open keyboard settings, or long-press the mic key. That opens the training screen.
- On the laptop, open
aegis-keyboardand press the voice button. That opens the training window.
The screen asks you to say a line. It scores the words the recognizer missed. You can teach a correction for a mishearing.
The slurred-speech switch
The switch is labeled “I’m drunk”. You turn it on yourself. It is for slurred speech. While it is on, the mic waits longer and uses only the corrections you teach in that state. It does not estimate a blood alcohol level. It does not detect that someone is inebriated. It is not listening in the background for drunkenness. Leave it off and dictation stays in the sober setting.
A correction learned in one state does not rewrite the other.
What leaves the device
The voice-profile text can be sent to the shared archive: the corrections and the scores. The audio is not sent. Because there are no separate accounts, that profile text is the profile the other connected copies can read. You are responsible for speech you train it on, including someone else’s.