You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Dictation is transcribed as one text. The keyboard used to cut a long dictation into pieces of under a minute, and sentences broke at the cuts ("..." and a new sentence in the middle of a thought). Now the whole recording is one request: with Offline Speech to Text 0.22 or newer it is transcribed in one go when you stop; older versions still get pieces. The text arrives when you stop, not while you speak.
Ready for live transcription: when the installed Offline Speech to Text can take a long live stream (proposed to its author, not in a released version yet), the audio is streamed while you speak and the text comes a few seconds after you stop. The recording is also kept in a file until the text is in, so a broken stream does not lose a dictation.
Words from your personal dictionary can be passed to Offline Speech to Text with each dictation, so that names are spelled right (also needs the proposed version). On by default, with a switch in Settings → Preferences → Dictation.
Rotating the phone no longer ends a dictation: recording goes on and the listening panel comes back. When the keyboard reappears while a dictation is being transcribed, the panel shows it, and the mic key shows it too.
A recording is limited to 30 minutes.
Setup wizard is easier to follow: only the button looks like a button, each step says what will open and what to do there, and it tells you that the list of keyboards may have to be scrolled to find Droneboard.