How it works, and why it is built this way.
Longer pieces about the decisions behind the app, and about dictation on the Mac generally. No announcements, no roadmaps.
-
Why Apple Dictation stops after 30 seconds
The setting that half fixes it, the two things that override the setting, and why a dictation session has a length at all.
-
The slow death of Dragon for Mac: a timeline
Dragon was the reason a lot of people could work on a Mac at all. Here is what happened to it, in order, and what the Mac has instead.
-
Why the speech engine ships inside the app
Zumbo is about 570 MB, which is large for a dictation app. That number is the whole product decision, so here is the reasoning behind it.
-
How on device dictation actually works on Apple silicon
Why a speech model that used to need a data centre now runs on a laptop, what the Neural Engine has to do with it, and where the remaining limits are.
-
What a vocabulary pack is and is not
A pack is not a list of words bolted onto the output. It changes what the recogniser is listening for, which is why it fixes things a find and replace cannot.
There is an RSS feed if you would rather not come back to check.