Superwhisper vs Apple Dictation: Which One I Actually Kept

I type for a living, more or less, and my wrists had started reminding me about it by mid-afternoon. Apple's own dictation was the obvious first fix — free, already there, zero setup — and it genuinely helped for short bursts. It's the long stuff where it fell apart: run-on paragraphs with no punctuation, and no memory of the product names and jargon I type every day. Superwhisper is the AI dictation app I installed to fix that specifically. A month later, I haven't uninstalled either one, and this is the honest split of which does which job.

This is beginner to intermediate — installing Superwhisper and dictating your first sentence takes about five minutes; building custom vocabulary and Modes is the part that rewards a bit more setup time. If you're weighing up a wider AI toolkit rather than just this one app, the full AI Tools guides and reviews is where I keep everything else I've actually tested.

What each one actually is

Apple's dictation is the free speech-to-text already built into every Mac and iPhone — tap the microphone key, or the dedicated shortcut in System Settings, and it types what you say directly into whatever text field has focus. Modern versions run mostly on-device, so it doesn't need an internet connection for most languages, and it understands basic spoken punctuation ("comma", "new line") if you remember to say it. It knows nothing about you beyond that: no custom spellings, no formatting preferences, no memory between sessions.

Superwhisper is a third-party app for Mac and iPhone that does the same basic job — turning your voice into text — using AI transcription models built on the same Whisper technology that powers a lot of today's speech recognition. The difference is everything around the transcription: a single global hotkey that works in any app, a choice between fast on-device models and larger, more accurate cloud models, custom vocabulary for names and jargon it would otherwise mangle, and "Modes" that reshape the output — tidying filler words for an email, keeping it verbatim for a quick note — depending on what you're dictating into.

Setting up Superwhisper: the five things that actually matter

1. Pick a hotkey you'll actually remember

Superwhisper defaults to holding a function key while you talk, then releasing it to transcribe — comfortable once it's muscle memory, but worth changing if it clashes with something else you already use. The whole point is that it works from any app without you opening Superwhisper itself, so the hotkey is the one setup decision that matters more than any other.

2. Choose your model: on-device speed vs cloud accuracy

This is the trade-off I didn't expect to care about as much as I do. The on-device models are fast and completely private — nothing leaves the Mac — but stumble slightly more on quick or accented speech. The cloud models are noticeably more accurate on exactly that, at the cost of a small delay and sending the audio off-device. I default to on-device for anything routine and switch to a cloud model deliberately for a long, important piece of dictation where I don't want to be editing typos afterward.

3. Build a custom vocabulary before you need it

Every job has words a generic model will never guess correctly — product names, people's names, industry shorthand. Superwhisper lets you add these to a personal dictionary so it stops guessing. I added "Britetechlifestyle," "Todoist" and a handful of client and product names in my first ten minutes, and it's the single change that made the difference between "impressively close" and "actually correct."

4. Set up Modes for the places you dictate most

A Mode tells Superwhisper how to reshape what you say for a given context — clean up an email into proper sentences and remove your "ums," or leave a quick note completely verbatim because you'll tidy it yourself later. I run one Mode for email and one for raw notes, and switching between them takes one keystroke rather than a settings menu.

5. Notice where you actually reach for it

In practice that turned out to be Slack replies, first drafts of emails, and — the one I didn't plan for — quick-adding tasks. If you've built any kind of capture habit around typing a task the second you think of it, dictating it instead removes the last bit of friction; it's exactly the kind of frictionless capture the Todoist task system I run depends on, and voice makes it faster still when my hands are already busy with something else.

Where Apple's dictation still wins

Zero setup is a genuine advantage, not a consolation prize. There's nothing to install, nothing to remember to update, and no separate subscription to think about — you tap the microphone and it works, on a device you already own. For a single sentence into a search box or a one-line reply, that's faster than Superwhisper could ever be, because there's no app-switching or hotkey to recall. It's also the only option that works identically well on the phone, where Superwhisper competes against a keyboard microphone that's already sitting right there with no extra tap required.

Where Superwhisper wins

Anything longer than a couple of sentences. Apple's dictation runs out of patience with itself on long-form speech — punctuation drifts, and there's no way to teach it your own vocabulary. Superwhisper's Modes and custom dictionary mean a three-paragraph email comes out closer to something I'd actually send, with far less cleanup afterward. The single global hotkey also matters more than it sounds: Apple's dictation behaves slightly differently depending on which app has focus, while Superwhisper works exactly the same everywhere, which is one less thing to think about mid-sentence.

The honest verdict: which one I actually kept

Common mistakes worth avoiding

1. Expecting clean formatting without setting up a Mode first. Superwhisper's raw transcription is accurate but unstyled — the formatting comes from telling it what context you're dictating into, not from the model guessing on its own.
2. Defaulting to the cloud model for everything. It's more accurate, but it's also the option that sends audio off your Mac and adds a small delay. Use it deliberately, not automatically.
3. Skipping custom vocabulary because "it'll probably guess right." It won't, consistently, for a name it's never seen — and fixing it once takes less time than correcting the same mistake every week.
4. Trying to replace every typing habit at once. The apps I dictate into changed gradually as I noticed friction, not because I sat down and decided to talk instead of type everywhere from day one.

If it's meetings rather than day-to-day typing that eat your time, this isn't the same tool for that job — dictation replaces your keyboard, but an AI notetaker listens to a call you're already having. I compared the actual options for that separately in Otter vs Fireflies vs Granola, and it's worth reading as a companion piece rather than a substitute for this one.

And if the honest bottleneck in your day is less "typing is slow" and more "I never trust which app to put the thing in afterward," that's a different problem with a different fix — I ran a proper month-long comparison in Todoist vs Apple Reminders: which one actually gets things done, and the two fixes ended up compounding rather than competing for my attention.

Whichever dictation app wins, what happens to the text afterward is a separate decision Apple's own operating system can now help with for free — I went feature by feature through the Apple Intelligence features actually worth turning on, and Writing Tools in particular pairs well with either dictation app above.

Questions people actually ask

Is Superwhisper free to try?

Yes — it installs with a free tier so you can test it in your own apps before paying anything. The free tier caps how much you can dictate each week; a paid plan removes that cap and unlocks the larger, more accurate transcription models. Try it on the free tier for at least a week before deciding whether the difference over Apple's dictation is worth paying for.

Does it actually work in every app, or just a few?

Every app, in my experience — Slack, Mail, Notes, a browser text box, a terminal. It works by typing the transcribed text wherever your cursor already is, the same way Apple's dictation does, rather than integrating with specific apps one at a time. If an app accepts a keyboard, Superwhisper can type into it.

Is my voice sent to a server, or processed on my Mac?

You choose. Superwhisper offers on-device transcription models that never leave your Mac, alongside larger cloud-based models that trade a little privacy for noticeably better accuracy on fast, mumbled or accented speech. I run the on-device model for anything sensitive and switch to the cloud model for long dictation where accuracy matters more.

Should I uninstall Apple's dictation once I have this?

No — I still use both. Apple's dictation needs no setup and no hotkey to remember, which makes it the right tool for a one-line reply you're typing right now. Superwhisper is what I reach for anything longer than a couple of sentences, where its formatting and custom vocabulary actually pay off.

Does Superwhisper work on iPhone as well as Mac?

Yes, there's an iOS app alongside the Mac app, using the same underlying models. I use it far less on the phone simply because Apple's own dictation is already built into the keyboard there with zero friction — the Mac is where a separate app earns its keep, because there's no equivalent built in.