Speech to text on a Mac: every option, and where each one stops
Turning what you say into typed text — not the other way round. A Mac gives you four separate ways to do it, they overlap confusingly, and each one hits a different wall.
Looking for the opposite direction?
This guide is about speech to text — you talk, your Mac types. If what you want is text to speech, your Mac reading a document out loud to you, that is a separate feature called Spoken Content, in System Settings under Accessibility, where you can also set a shortcut that speaks whatever is selected.
Every Mac can already do this. You do not need to install anything to dictate a sentence into an email, and if that is all you came for, the built-in feature will not disappoint you. What tends to disappoint people is the second week: the session that cuts out mid-thought, the colleague's name that comes back spelled three different ways, the punctuation you have to say out loud, the fact that the trick you learned in Pages does not work the same way in your browser.
That happens because “speech to text on a Mac” is not one thing. It is at least four, built at different times by different people, and the reason people bounce between them is that nobody ever explains which is which. This guide does that first, then says plainly where each one runs out.
The four ways a Mac turns speech into text
Read the last column first. Every one of these works; the only question worth asking is which wall you are going to hit and whether you care.
| Option | Best for | Where it lives | Where it stops |
|---|---|---|---|
| Built-in dictation | Typing a sentence or a paragraph into whatever app is already open | System Settings → Keyboard, in the Dictation section | You say the punctuation out loud, the session ends by itself, and there is no way to teach it a name it keeps getting wrong. |
| Voice Control | Running the whole machine by voice — clicking, scrolling, editing, dictating | System Settings → Accessibility, under Voice Control | It listens for commands the entire time it is on, which is exactly what you want for hands-free use and exactly what you do not want when you only meant to type a line. |
| An app's own dictate button | Long-form writing inside the one app that has a good implementation | Inside the app — Word's Dictate button, Google Docs' Voice typing under Tools | It is a different engine with different accuracy and different rules in every app, and it does nothing at all in the next window you click into. |
| A third-party dictation app | One consistent shortcut that behaves the same way everywhere | Runs as a menu-bar app, with microphone and accessibility permission | It costs money, and it is another background app on a machine that already has a feature that is free. |
Table scrolls sideways on a narrow screen.
Built-in dictation: what it actually is
Apple's dictation is a system-level feature, not an app. You switch it on once in System Settings, under Keyboard, in the section marked Dictation — on older versions of macOS the same controls live in System Preferences under Keyboard. From then on it is available in any standard text field on the machine. Put the cursor where you want the words, trigger dictation, talk, and the text is typed in at the insertion point exactly as though it came from the keyboard.
That is the important part, and it is the part people underestimate: this is not a window you dictate into and then copy out of. It targets the cursor. A search field, a Finder rename, a comment box, a Slack message and a Pages document are all the same case as far as the system is concerned.
Punctuation is the first thing that surprises people. You say it: “comma”, “period”, “new paragraph”. Recent versions of macOS add some punctuation on their own in supported languages, but the reliable way to get a comma where you want one is still to ask for it out loud. It works, and after a few hundred sentences you stop noticing you are doing it — but you are doing it.
Where your speech is processed depends on your Mac and your language. On Apple silicon with a supported language, dictation runs on the device: it works with no network, and there is no session length to run into. Otherwise audio is sent to Apple's servers for recognition, which is why two Macs sitting on the same desk can behave differently — one dictates for as long as you like, the other stops on its own after a stretch. Neither is broken; they are running different engines.
Turning it on, the full punctuation command list, and the fixes for when it stops responding are all covered click by click in the blog post below. This guide deliberately does not repeat them.
Voice Control is a different feature, and half the people reading this want it
Voice Control lives in System Settings under Accessibility, and it is a much larger thing than dictation. It is built to run a Mac entirely by voice: clicking, scrolling, switching apps, selecting and correcting text, and dictating into it. It ships with a command vocabulary you can see and extend, it can put numbers or a grid over the screen so you can refer to things you cannot name, and once its files have downloaded it does its work on the machine.
If you are looking at this page because typing hurts, or because you cannot use a keyboard reliably, Voice Control is the feature you want and it is worth the afternoon it takes to learn properly. It is genuinely good, and it is free.
It is a poor fit for the other case. Because it is always listening for commands while it is on, ordinary phrases occasionally get taken as instructions, and you spend attention on staying out of its way. For someone who just wants to say two sentences into a reply and get back to work, that is the wrong trade.
The dictate buttons inside individual apps
Several apps ship their own dictation, unrelated to the system's. Microsoft Word for Mac has a Dictate button on the Home tab. Google Docs has Voice typing in the Tools menu, which works in Chrome and only inside the document body. A growing number of web apps put a microphone in a corner of the compose box.
Each of these is a separate engine with its own accuracy, its own punctuation rules and its own idea of when it has stopped listening. They can be very good inside their own window — Word's, in particular, is better than people expect — but they share one flaw, which is that everything you learned stops working the moment you click into another app.
This is the quiet reason most people give up on dictating. Not that any single tool was bad, but that using three of them means never getting fluent at any of them.
Transcribing a recording is a different job entirely
None of the above will take an audio file you already have. Dictation and Voice Control listen to a microphone in real time and type as you go; they have no concept of a file on disk. If what you need is a transcript of a meeting recording, a voice memo or an interview, that is a different category of tool, and the ones built for it are built around timestamps, speaker labels and editing a transcript after the fact.
Apple has been adding transcription to its own recording apps, and there are good dedicated services for it. The point is only that live dictation and file transcription are separate problems: the tool that is best at one is usually not the tool you want for the other, and searching for one and buying the other is a common and expensive mistake.
Where the built-in options stop being enough
This is the honest part, and it cuts both ways. For a great many people the answer is that they never stop being enough. If you dictate a few sentences a week into Notes, macOS already does that for free and there is nothing here worth paying for.
The complaints that do come up are consistent, and they are about the shape of the feature rather than the quality of the recognition:
- You have to speak the punctuation. Every comma, every full stop, every paragraph break is a word you say out loud, and it changes how you think while you talk.
- The session has a mind of its own. Server-based dictation ends after a stretch, and silence ends it early, so long-form dictation becomes a series of restarts.
- It cannot be taught. There is no vocabulary list to hand it your colleagues' names, your product's spelling or the acronyms your team uses, so the same three words come back wrong forever.
- It transcribes exactly what you said, filler and all. “Um”, “you know” and the false start you abandoned halfway all get typed, and cleaning that up costs more than the dictation saved.
- Nothing is kept. The text lands in the field and that is the end of it. There is no record of what you dictated, so a paragraph you got right the first time cannot be found again.
- The experience differs by app and by machine, which makes it hard to trust. Something that works in one window and not the next never becomes a habit.
So which one should you use
If you dictate occasionally and mostly into Apple's own apps, use the built-in feature and stop reading — it is already installed and it is free.
If you need to operate the machine without a keyboard, learn Voice Control properly. Nothing you can buy replaces what it does, and it is designed with that need at its centre.
If you write long documents inside one app that has a good dictate button, use that button in that app and do not try to make it your everything.
And if dictation is how you want to write — most of your email, most of your messages, most of your notes, across a dozen apps in a day — that is the case the built-in options were never designed for, and it is the case a dedicated app is for.
Where VoiceSnap Pro fits
VoiceSnap Pro is built for the last of those four cases: one held shortcut that behaves identically in every app on the machine, punctuation inferred from how you speak rather than dictated word by word, and a searchable library that keeps everything you said instead of letting it evaporate into a text field.
It is not shipping yet. There is no download, no trial and no reviews; the download page is a waitlist and nothing else. When it does ship it is a one-time purchase rather than a subscription, which is the whole point of it. If Apple's dictation already covers what you do, the honest advice is to keep using it.
What it does differently
- One held shortcut that behaves the same way in every app on the machine — no per-app engine, no browser extension, no window to copy out of.
- Punctuation, capitals and paragraph breaks inferred from how you speak, so you never say the word “comma” out loud.
- Filler words and false starts removed on the way in, rather than cleaned up afterwards.
- A custom vocabulary for the names, acronyms and jargon a general model keeps guessing at.
- Every dictation saved to a searchable notes library instead of vanishing into the field it landed in.
- A one-time purchase. No subscription, and nothing you dictate is used to train AI models.
None of this has shipped. There is no download, no trial, no customers and no reviews — the download page collects an email address and nothing else.
Frequently asked questions
No — they are opposites, and the search results for both are hopelessly tangled. Speech to text listens to you and types what you said; that is dictation, and it is what this guide is about. Text to speech reads what is on screen out loud; on a Mac that is Spoken Content, in System Settings under Accessibility, and it has its own speak-selection shortcut.
Yes. Dictation is part of macOS and has been for many years. You switch it on in System Settings under Keyboard, in the Dictation section, and after that it works in any standard text field on the machine. Nothing to install and nothing to pay.
It depends on the Mac and the language. On Apple silicon with a supported language, dictation runs on the device and works with no network at all. Otherwise the audio is sent to Apple for recognition, which needs a connection. The Dictation section of Keyboard settings is where you see which languages you have available.
Usually because it is running the server-based engine, which works in limited stretches rather than indefinitely, and because a pause of a few seconds is treated as the end of what you were saying. On-device dictation does not have the same limit. The blog post linked from this guide walks through the specific fixes.
The built-in feature works in any standard text field, which covers the overwhelming majority of apps, because it types at the cursor rather than integrating with anything. Apps that draw their own text engine — some editors, some games, some remote-desktop windows — can behave unpredictably, and that is true of any dictation tool, not just Apple's.
For ordinary sentences, yes — recognition is not usually the weak point. It struggles in the predictable places: proper names, industry jargon, acronyms, and anything where you needed punctuation you forgot to say. Those are the things a tool with a custom vocabulary and automatic punctuation is built to fix.
Everything Apple builds in is free and already on your Mac. Third-party dictation apps charge — most of them monthly, which is worth knowing before you start a trial. VoiceSnap Pro will be a one-time purchase, and it has not shipped yet.
Related reading
The step-by-step articles, the comparisons and the free tools that go with this guide.
Get told once, on release day
VoiceSnap Pro is a one-time purchase with no subscription, and it is not out yet. One email when it is — nothing else.
One launch email. No newsletter, no sharing your address. Privacy policy