How to Dictate Into Any Mac App With a Shortcut

What has to be true for this to work

System-wide dictation on a Mac is four moving parts. When it fails, one of the four is missing, and knowing the list saves you from reinstalling an app that was never broken.

1. A global shortcut the app has registered with the system.
2. Microphone permission, so it can hear you.
3. Accessibility permission, so it can put text into a different app.
4. A focused text field with a blinking caret in it.

Apple Dictation handles the first three internally with Fn or the microphone key. Third-party apps ask you for two of them explicitly. Auri Voice follows the same pattern on macOS 15 and later, with a shortcut you choose and local engines doing the recognition.

Pick a shortcut that does not fight your Mac

The shortcut is the part you will touch fifty times a day, so spend a minute on it. It should be reachable without looking, and it should not already belong to something else. Left Function is the most common pick for a reason: it is close, it is single-key, and most people never use it for anything.

Good candidates and things to avoid:

• Good: Left Function, Right Option, a spare function key, or a side button on your mouse.
• Avoid: anything with Command and Shift that a screenshot tool or window manager might claim.
• Avoid: Fn double press if you already use Apple Dictation, so the two do not race each other.

Then choose how it behaves. Toggle means press to start, press to stop, which is what Auri Voice uses by default and what suits longer drafting. Hold-to-talk means it listens only while the key is down, which is safer in an open office because you cannot walk away from a live mic.

After you set it, trigger it once from the desktop. If Siri or the emoji picker appears, change it immediately rather than training your hands on a conflict.

Grant Microphone and Accessibility

These two permissions cause more support questions than anything else, because the failure looks like broken recognition. The app hears you fine, shows a transcript in its own window, and nothing arrives in Mail. That is Accessibility, not the microphone and not the model.

1. Open System Settings, then Privacy & Security, then Microphone. Enable your dictation app.
2. Go to Accessibility in the same section. Enable it there too.
3. Quit and reopen the app if macOS asks.
4. Trigger the shortcut once and accept any follow-up prompt.

If insertion still fails, remove the app from the Accessibility list, restart the Mac, and add it back. That sequence clears a stuck permission state that macOS otherwise keeps insisting is fine.

Recheck both after major macOS updates. They reset more often than you would expect, usually on the morning you have something to send.

Test in Notes, Mail, and Slack

Test in three apps, in this order, using the same spoken sentence each time. Changing what you say while also changing the app makes it impossible to tell what went wrong.

1. Notes: the simplest case. Click into a note, trigger, say a full sentence, stop. Confirms the basic path works.
2. Mail: a longer compose window. Dictate two sentences and a line break. Confirms paragraph handling.
3. Slack: the hard one. Click the message box in a DM, then dictate a short list. Slack has several panes that can hold focus, and it is the app where insertion most often lands nowhere.

Add a browser test if you write in Gmail or Google Docs. Web editors sometimes hand focus to a notification or an autosave banner mid-sentence.

If Notes works and Slack does not, the app is fine. It is focus. Click the message field again and retry before you start changing engines.

Fix focus and cursor problems

Almost every "dictation is broken" report is a focus or permission problem wearing a costume. Run through this list before touching any setting related to speech.

• Nothing inserted: the caret was not blinking. Click into the field and trigger again.
• Text in the wrong app: you switched apps while speaking, or a notification stole focus. Keep the target window in front until insertion finishes.
• Shortcut opens Siri or emoji: the hotkey collides with Fn or Globe behavior. Pick a different key.
• Works once after a reboot, then stops: Accessibility permission was cleared by an update. Toggle it off and on.
• Text arrives scrambled or in the wrong place: a clipboard manager or text expander is inserting at the same moment. Quit it and retest.
• First word missing: you started speaking before the app started listening. Wait for the indicator.
• Works everywhere except one app: some surfaces refuse inserted text, including secure password fields, virtual machines, and remote desktop sessions. Dictate into a note and paste across.

Thirty seconds through this list beats an hour of downloading different speech models to solve a problem that was never about speech.

Why cleanup matters once the text lands

Getting a shortcut working is the easy half. The half that decides whether you keep using it is what arrives in the field. Raw speech looks bad in writing, even when the transcription is perfect, because people do not talk in finished sentences.

Here is the same dictation, two ways. Spoken: "hey can you, uh, take a look at the deck before, before the call, actually before end of day would be better, thanks". As a raw transcript, that is what you get, and you now edit a Slack message in front of a colleague. With a cleanup pass: "Can you take a look at the deck before end of day? Thanks."

That is the difference between a shortcut you use twice and a shortcut you use all day. Auri Voice runs that cleanup on your Mac by default: filler removed, punctuation placed, mid-sentence corrections resolved to your final intent, so the text at the cursor is already sendable.

Once you trust it, your speaking changes. You stop composing carefully, you think out loud, and you go roughly five times faster than typing. That only works if you are not proofreading every insert.

FAQ