How to use dictation on your Mac

What macOS Dictation actually is

macOS Dictation is a built-in feature that lets you speak instead of type, anywhere you can enter text on your Mac. It ships with the operating system at no extra cost. You don't install anything. You turn it on once in System Settings, and from then on it's available in every app — a Mail draft, a Notes page, a search field, a comment box on a website.

It isn't a transcription app for audio files, and it isn't a meeting recorder. It's live speech-to-text: you talk, words appear at the cursor in real time. Apple's documentation lists support across more than 40 language and region combinations, so it isn't English-only. On a Mac with Apple Silicon, general text dictation can be processed on your device rather than sent to Apple's servers, which is the part worth caring about if you ever dictate anything you'd rather keep off a server.

One honest caveat up front. The built-in feature is genuinely good at what it's designed for — short, everyday dictation. It gets less reliable the longer you go and the messier your audio. That's not a knock; it's a design boundary. Knowing where that boundary sits is the whole point of this guide, and I'll mark it clearly before the end.

Turn it on in System Settings, then just talk

The switch lives in one place. Open the Apple menu in the top-left corner, choose System Settings, then click Keyboard in the sidebar (you may have to scroll down to find it). Look for the Dictation section and turn it on. macOS will ask you to confirm with an Enable prompt, and it may ask whether you want to share audio recordings with Apple — that part is optional, and you can decline.

Once it's on, you start dictation one of three ways, per Apple's own instructions: press the Microphone key if your keyboard has one in the function row, use the Dictation keyboard shortcut, or choose Edit then Start Dictation from the menu bar. The shortcut isn't fixed — in the Dictation settings you can open the Shortcut menu and pick one, or choose Customize and press the keys you want. So "the shortcut" is whatever you set it to, which is the first thing I'd change if the default ever clashes with something else you use.

The flow itself is simple. Click into any text field so the cursor is blinking, trigger dictation, wait for the cursor to start pulsing, and speak. On Apple Silicon you can keep using the keyboard while you talk — no need to stop dictation to fix a word. When you're done, you trigger the shortcut again or stop in the menu. That's the entire built-in feature: one toggle, one shortcut, every text box on the machine.

If dictation cuts out partway through a sentence, or refuses to start, that's usually a permissions or microphone hiccup rather than something you did wrong — the fixes for Mac dictation that keeps stopping cover the common ones.

The other way: one push-to-talk hotkey

A dedicated tool works on the same principle as the built-in — press a key, speak, text lands at your cursor — but it doesn't quietly stop on long passages, it lets you teach it words it keeps getting wrong, and it can clean up the result before it pastes. Whisper is one of these. You need a Mac on Apple Silicon, a working microphone, and a couple of minutes. The whole local pipeline is free for any signed-in account, with no card asked for at sign-up. Here's the sequence.

The slow part is the one-time model download, not the setup. Everything after that is the four steps above. The mechanic is the same muscle memory as the built-in feature — the difference is what happens on the long, messy passages where macOS Dictation starts to wobble.

Punctuation, commands, and which languages work

With macOS Dictation, punctuation comes from voice commands. You say the name of the mark and it appears — "period," "comma," "question mark," "exclamation mark." You can say "new line" or "new paragraph" to break text, and even spell out emoji by name, like "heart emoji." In supported languages, macOS can also add commas and periods automatically; there's a toggle for automatic punctuation in the Dictation settings if you'd rather it guess than be told.

A dedicated tool handles punctuation differently. Instead of you narrating every comma, it can punctuate the whole passage in one cleanup pass after you stop speaking, which I find far less exhausting over a long block — saying "comma" two hundred times a day gets old. It also handles a problem the built-in can't: words it keeps mishearing. Whisper has a custom-vocabulary feature, so the names, product terms, and acronyms you use constantly stop coming out as nonsense. macOS Dictation has no equivalent dictionary you can feed.

On languages, both cover more than English. Apple lists 40-plus language and region variants for Dictation. Whisper's multilingual models go to 99 languages and can translate to English; its faster Parakeet engine covers English plus 24 European languages. So if you write in two languages, or you want translation built in, that's a point in the dedicated tool's column — but if you only ever dictate in English, the gap closes a lot.

Local or cloud: keeping it on your Mac

macOS Dictation already leans local on Apple Silicon — general text can be processed on-device, which is one of its genuine strengths. A dedicated tool gives you the same on-device option, plus a cloud path for when you want top-tier accuracy. Here's how the three paths differ, because a good tool makes you choose rather than choosing for you:

• Local Parakeet — NVIDIA's TDT engine, around 600 MB, and the fastest local option — 5 to 10 times faster than Whisper on CPU. Covers English plus 24 other European languages, 25 in total. No translate-to-English. If you dictate in English or another European language, this is the quick, fully offline pick.
• Local Whisper — slower than Parakeet on the same machine, but the multilingual builds cover 99 languages and can translate to English. The English-only builds are English-only, not 99. Pick this for Chinese, Japanese, Korean, or any translation work, which Parakeet can't do. Default English model is around 480 MB.
• Cloud (OpenAI, BYOK) — best accuracy and web access, using your own OpenAI key billed straight by OpenAI. Transcription runs on gpt-4o-mini-transcribe by default. Needs internet, so it's the one path that leaves your machine. The Cloud surface is part of Whisper Pro.

The boring truth is that for most of what people dictate on a Mac — emails, notes, messages — local is plenty, and both local engines run with nothing sent to a server. Cloud earns its place when you want the highest accuracy on a hard recording or you need the model to pull a fact off the web mid-sentence. For everyday dictation, start local and only reach for cloud when local leaves you wanting. If you mostly live in one app, the same logic plays out in voice-to-text in Apple Notes, where the cursor — not the app — is what does the work.

Getting clean text, not a run-on

Raw dictation comes out as a wall of words. You say "okay so reply to the landlord about the leak ask when the plumber can come and cc my wife," and that's the unpunctuated run-on any speech engine hands you first. With macOS Dictation you fix that live, by narrating the commas yourself or letting auto-punctuation guess. It works, but you're doing the editing in real time, mid-sentence.

A dedicated tool can do the cleanup after you stop talking. Whisper can run an AI pass over the raw text — stripping the "ums," fixing the run-ons, adding punctuation — before a single word lands at your cursor. You trigger it with the activation phrase "Hey whisper." On a local model that pass runs through Ollama on your machine; in cloud mode it's gpt-5-mini by default. Either way, what arrives is the tidied version, not the raw one.

This is the difference that shows up over a workday, not in a single sentence. For one short message, narrating your own punctuation is fine. For a dozen of them, or a long note, having the tool clean the whole passage in one go is the part you stop wanting to live without. The same speak-then-clean flow is what makes it practical to write faster by voice across every app, not just dictate the occasional line.

When macOS Dictation is all you need

Sometimes the built-in feature is the right call, and pretending otherwise would be dishonest. If you mostly dictate short things — a one-line text, a quick search, a 30-word reply — macOS Dictation covers it for free, on-device on Apple Silicon, with nothing to install. Turn it on in System Settings, set a shortcut you like, and you're done. I'm not going to tell you to add an app for a sentence.

I'd reach for a dedicated tool when the built-in starts to hurt: long passages where it stops on you, words it keeps mishearing that you'd like to teach it once, narrating punctuation a hundred times a day, or wanting one cleanup pass instead of live editing. That's roughly the point where saying "comma, period, new paragraph" out loud costs more attention than it saves. Below that bar, the free Apple feature wins, and you should use it.

There's also a hardware truth that sits under all of this. A $20 USB microphone does more for accuracy than any setting, on the built-in feature or a dedicated one. If your dictation is coming out garbled, fix the mic before you blame the software. And if macOS Dictation specifically keeps quitting on you, that's usually a fixable system issue — sometimes it's a runaway corespeechd process eating CPU, not the feature being broken.

The whole thing comes down to one toggle and one honest line. macOS Dictation has been on every Mac for years, it's free, and for short bursts it's exactly right. The moment you're writing for real — long, fast, in your own vocabulary — a tool built for that takes over. I dictated most of this guide with the hotkey, narrating zero commas, while a cleanup pass handled the punctuation I'd otherwise have said out loud two hundred times. My family would tell you I talk enough as it is.