Voice dictation
A mic button in the composer turns speech into text in real time. It is built on the WebView’s native SpeechRecognition API, so there is no extra model to install — but it also means the platform itself decides what happens to the audio.
How it works
Click the mic icon to start recording. The current contents of the composer are snapshotted as your “base text”; interim guesses render as a preview appended to the base while you speak, and final phrases get committed as soon as the engine settles on them. You can keep typing between phrases — mid-sentence updates won’t clobber your edits, because they always replay over the same anchored base text.
Click the mic again to stop. Submitting the message auto-stops the recogniser for you. Benign errors like “no speech detected” or “aborted” are swallowed silently; a permission denial surfaces a short hint so you can fix it in browser settings.
Where it shows up
Voice dictation depends on the platform exposing the Web Speech API. In practice:
- Windows — supported via WebView2.
- Linux — usually unavailable. The mic button is hidden on systems whose WebView does not expose the API rather than showing a button that does nothing.
On Chromium-based engines, audio for the speech recogniser is typically routed through Google’s hosted speech service. Voice dictation is therefore not strictly offline, even though the rest of Loach is local-first. If that is a concern, stick to typing.