Control your Mac with your voice: Voice Control or Yubi
Apple's Voice Control reaches every button by voice and works offline. How to turn it on, the commands worth learning, and where Yubi fits instead.
The feature that lets you run a Mac entirely by voice is called Voice Control. It has shipped with macOS since Catalina, and it lives in the Accessibility settings, which are easy to miss if you never open them.
I dictate all day in Tokyo and I build Yubi, so I have spent a lot of time with both. They are good at different things.
Turning it on
Open System Settings from the Apple menu, click Accessibility in the sidebar, then click Voice Control and switch it on. The first time, your Mac needs an internet connection for a one time download from Apple. After that download, Voice Control works without a connection.
Apple's 2019 technical brief on Voice Control says its audio processing happens on the device, which is why it keeps working offline.
The commands worth learning
Voice Control handles dictation and commands in the same stream. Talk, and it types. Say a command, and it acts. Apple's own example: in Messages, say "Happy birthday. Tap send." and only "Happy birthday" is sent.
For anything on screen, three commands do most of the work. "Show names" labels the items you can click, so you can say the name. "Show numbers" puts a number on each item instead, which helps with unlabeled buttons or a row of photos. "Show grid" lays a numbered grid over the screen, and saying a number zooms into a finer grid in that spot. "Show window grid" keeps it to the front window.
When you forget a command, say "Show commands" to see the ones for the app you are in. The full list sits in System Settings, under Accessibility, Voice Control, Commands.
You can add your own commands and vocabulary too. If you say client names or product terms every day, add them early. Vocabulary is not available in every language, and neither is spelling mode.
Who it is built for
Voice Control was built so people who cannot use a keyboard or mouse can run the entire Mac by voice. For that job it is the right tool, and it is free. Everything you can click has a command, and the grid reaches places nothing on screen names.
If your hands are free and you just want to get something done faster, it can feel slow. You drive the Mac one step at a time. To change a setting you have never touched, you still need to know where it lives, and then you say every click on the way there.
Where Yubi fits
That second case is the one I built Yubi for. Hold the right Option key and ask something like "How do I open a new tab in Chrome". She reads the window's accessibility tree, the same structure VoiceOver relies on, finds the button and points at it. By default you do the clicking. Turn on one switch and she presses it for you, with the same accessibility action VoiceOver uses.
She reads labels, not pixels, so she never asks for Screen Recording. Asking about the screen is optional and uses one task from your plan. Dictation stays on the Mac and costs nothing. Yubi needs macOS 26 or later and an Apple Silicon Mac.
Some apps draw their own canvas and expose almost nothing to read. There, Yubi tells you she cannot see the window instead of guessing. Voice Control's grid still works in those apps, because a grid square is only a position on the screen.
Which one to use
If you need the whole Mac by voice, turn on Voice Control and start with the three show commands. If you mostly type and want your voice for dictation and the occasional how do I, Yubi asks less of you: hold a key and talk.