Speech to text on Mac: how to turn it on and actually use it
Mac has had speech to text built in for years, but the setup screen does not tell you the shortcut, which voice commands are real, or where your words actually go.
Short answer: Turn on speech to text on a Mac in System Settings, Keyboard, Dictation, then press the microphone or Globe key (or tap Fn twice on older keyboards) in any text field to start talking. On an Apple Silicon Mac it transcribes on your machine once the language file has downloaded; typing stops after about 30 seconds of silence.
How to turn on speech to text on a Mac
This is built into macOS and free. No app to install for the basic version.
- Open the Apple menu, then System Settings.
- Click Keyboard in the sidebar.
- Find Dictation and turn it on. Click Enable if macOS asks.
- Pick a language, and choose whether to share audio recordings with Apple to help improve the feature (this is optional and separate from whether your dictation itself is processed locally).
- Click into any text field, press the shortcut below, and start talking.
That is the whole setup, per Apple's own support page. It works in Mail, Messages, Notes, a browser text box, almost anywhere you can place a cursor.
The keyboard shortcut for Mac speech to text
On a keyboard with a dedicated microphone or Globe key, press that key once to start. On older Mac keyboards, the default is pressing Fn twice. Both start and stop dictation, and you can set a different shortcut under Keyboard, Dictation in System Settings if the default clashes with something else you use.
One thing that surprises people coming from a hold-to-talk tool: the built-in shortcut starts an always-listening session, not a hold-to-talk one. Tap it, talk, tap it again (or wait out the silence timeout) to stop. There is no key to hold down.
Dictation keeps listening for as long as you keep talking, but it stops itself after about 30 seconds without any speech, so a session you forgot about will not run forever.
Punctuation and editing: what is real and what is not
Say the name of a punctuation mark and Apple's dictation types the symbol: "comma," "question mark," "exclamation mark." Say "new line" for a single line break or "new paragraph" for a full paragraph break. In supported languages, Dictation also auto-inserts commas and periods for you as you talk, without you saying anything.
What is not part of Dictation, despite showing up in a lot of "voice commands" roundups: things like "select that word" or "delete last sentence." Those belong to Voice Control, a separate accessibility feature that lets you operate the whole Mac by voice, mouse clicks and all. Turning on Dictation does not turn on Voice Control, and the two have different command sets. If you want hands-free editing, not just hands-free typing, Voice Control (under Accessibility in System Settings) is the one you actually want.
Does it send your voice to Apple, or keep it on your Mac?
Both exist, and which one you get depends on the hardware. On an Apple Silicon Mac (M1 or later), Dictation runs on your machine once its language file has finished downloading the first time you use it, so your audio does not need to leave the computer for ordinary dictation. In Keyboard settings you can check whether your voice input is being processed on-device or sent to Siri servers. Apple has also said that for some requests the on-device model cannot handle with confidence, audio may still be routed to their servers, and older Intel Macs need an internet connection because they rely on server-side processing by default.
The separate "Improve Siri & Dictation" toggle, which asks whether Apple can store audio samples to improve the feature, is opt-in and is not the same switch as on-device versus cloud processing. You can decline it and still use dictation normally.
Why this matters beyond privacy: on-device processing is also why dictation on an Apple Silicon Mac keeps working on a plane or anywhere without a signal, once the language file is already downloaded, while an older Intel Mac dictating over a slow connection will feel laggy or drop words entirely.
The real limits of Apple dictation
An honest list, because the setup page will not tell you this part:
- It is not available in every language or region. Coverage is decent but not universal, and features can vary by language.
- It stops after 30 seconds of silence. Fine for a quick sentence, annoying if you pause to think mid-thought.
- No custom vocabulary. A product name, a person's name, or a technical term you use daily gets no special treatment, and there is no way to teach it one.
- It only ever transcribes live speech into whatever text field has focus. It does not accept a file, so an existing recording is a different job entirely, covered next.
Transcribing an audio file is a different job
Dictation only works on speech happening right now, into a text field that is focused. If you already have a recording, a voice memo, an interview, a meeting someone else recorded, Dictation cannot touch it. For that you need something that accepts a file:
- On macOS Sequoia and later, Notes and Voice Memos can import or record audio and transcribe it, then let you copy the text out. It is built in and free but limited to those two apps.
- For batch transcription, more formats, and more control (speaker separation, export formats, longer files), a dedicated tool like MacWhisper is built specifically for feeding it existing audio and video files rather than for talking live.
A free way to talk into any app, not just one text field
If what you actually want is to talk instead of type across everything you use all day, Slack, your terminal, an email, a code editor, Apple's Dictation gets you most of the way but stops at its own limits above. MightyMouse turns talking into typing in any Mac app, using local Whisper transcription that runs on your own machine the same way Apple's on-device mode does, and it is free: connect the Claude, Codex, Gemini, or OpenCode account you already have, or run a local model, and there is nothing to pay for the transcription itself.
Two dictation styles cover the push-to-talk and no-hands-free cases: hold a key and let go when you are done, or double-tap to start hands-free dictation that keeps typing as you pause and keep talking, without a key held down. Both use the same local transcription, they just differ in how a sentence ends.
It is not a replacement for Voice Control's hands-free editing commands, and it does not import audio files either, that is still a job for Notes or MacWhisper. What it replaces is the ceiling on where Apple's dictation works: one text field at a time, with whatever app happens to have focus, rather than everywhere you type.
Speech to text on a MacBook vs a MacBook Pro vs a MacBook Air
The setup steps and shortcut are identical across every current Mac, MacBook Air, MacBook Pro, iMac, Mac mini, Mac Studio. The only thing that changes is whether dictation runs on-device: any Apple Silicon Mac (M1 and later, which is every MacBook Air and MacBook Pro sold since late 2020) gets local processing. An older Intel MacBook Pro falls back to sending audio to Apple's servers and needs an internet connection to dictate at all. If you are shopping and speech to text is a deciding factor, the chip generation matters more than which MacBook line you pick.
Talk instead of typing, in any Mac app.
Free local transcription. No card to try it.
Common questions
How do I turn on speech to text on my Mac?
Open System Settings, click Keyboard, then turn on Dictation under the Dictation section and click Enable if asked. After that, press the microphone or Globe key (or Fn twice on older keyboards) in any text field to start talking.
What is the keyboard shortcut for dictation on a Mac?
The microphone or Globe key on newer keyboards, or pressing Fn twice on older ones. You can set a different shortcut under Keyboard, Dictation in System Settings.
Does Mac dictation work offline?
On an Apple Silicon Mac (M1 or later), yes, once the language file has downloaded once, dictation processes on your machine without needing the internet. Older Intel Macs generally need a connection because they send audio to Apple's servers.
Why did my dictation stop by itself?
Apple's Dictation stops automatically after about 30 seconds without any speech. That is normal behavior, not a bug, and starting it again just means pressing the shortcut once more.
Can I use Mac dictation to transcribe a recording I already made?
No, Dictation only converts live speech into whatever text field has focus. To transcribe an existing audio file, use Notes or Voice Memos on macOS Sequoia and later, or a dedicated tool built for files like MacWhisper.
Once you have hit the limits above, MightyMouse turns talking into typing in any Mac app, with free local transcription and no monthly word cap.