Start with how many Macs you own

This comparison has an unusually short arithmetic layer, so it may as well go first. From VoiceInk's pricing page on 4 September 2026: Solo is "Now $25", for one Mac. Personal is "Now $39", for up to two. Extended is "Now $49", for up to three. All three are described as "One-time purchase, lifetime access" with a "14-day money-back guarantee".

Coii VoiceInput is $19 once and the licence covers three devices. There is one tier and no larger one.

On one machine it is a $6 difference and nobody should choose software over $6. On three it is $19 against $49, which is a real gap — and it is also the point where somebody with a desktop, a laptop and an old machine in the spare room starts paying attention. Neither of these is expensive. They are just priced on different axes, and yours might be the axis where it matters.

Everything else on this page is more interesting than the money.

The requirement that ends the discussion for some readers

VoiceInk's pricing page states the requirement plainly: "Works only on Apple Silicon Macs · Requires macOS 14.4 or later". Their homepage repeats it — "VoiceInk requires Apple Silicon and macOS 14.4 or later" — and recommends 8GB of RAM.

Coii VoiceInput asks for macOS 13 Ventura or later and runs on Apple Silicon and Intel.

If you are on a 2019 MacBook Pro, a 2018 mini, or any Mac that never got Apple silicon, this is not a comparison at all — one of the two will not install. The same is true a step down: a Mac held at Ventura by a managed update policy is below VoiceInk's floor and above ours. This is the least glamorous section on the page and it decides more purchases than the feature list does.

Where VoiceInk is the better product

Feature for feature it is not close, and the page would be worthless if it pretended otherwise. From their own site:

It is open source. "VoiceInk is built in public on GitHub. Every line is out there for you to read, audit, or run yourself." That is a claim this app cannot make in any form. If auditability is what you are shopping for, stop here.

There is an iOS app. Their site says it is "Now available on iOS". This app is macOS and nothing else — no iPhone, no iPad, no Windows, no Linux. If part of your writing happens on a phone, they have a story and we have none.

A custom dictionary. VoiceInk advertises custom word recognition and smart phrase replacement. Coii VoiceInput does not learn your jargon, your colleagues' surnames or your product names, and never will — it has no vocabulary layer at all. Over months, that is the difference people notice most.

Modes and rewriting. Writing modes for email, notes and list formatting, plus a text-rewriting capability and AI-powered enhancement. You can ask VoiceInk to shape the output. You cannot ask this app anything; it types what you said.

A journal with timestamps. They describe journal integration. This app keeps a history of dictations, which is a different and smaller thing.

Cloud Enhancement, if you want it. Optional, and they are precise about the boundary: "only transcribed text (not your voice) is processed if you choose to enable it." An opt-in that is clearly labelled is good practice and worth saying so.

That is five capabilities and a platform. Against that, this app has four behaviours and a price.

What this one does instead

The pitch is not "more features". It is that there is one behaviour and it is built so that the behaviour cannot cost you anything.

The transcript exists before the engine is consulted

Release the key and the row is written to your local history first. Only then does the rest of the sequence run. A slow engine, a quit application, a window that closed, a paste an application declines to accept — all of it happens after the words are already on disk, so the worst case is retrieving them rather than saying them again.

The destination is fixed when the key goes down

Not when you release it. Start dictating into a comment field, look away, come back, and the text is in the comment field. This is the failure that most often goes unnoticed with other tools, because the words do appear — just somewhere else, sometimes in something that was listening for keystrokes.

The clipboard survives the paste

Where an application does not expose its field through the accessibility route, insertion is a synthetic paste, and whatever you had copied is restored afterwards.

One price and no second decision

$19, three devices, updates included, nothing renews. The trial is thirty days of the entire app with no card and no account, which is longer than VoiceInk's 14-day money-back window and, more importantly, is a trial rather than a refund — you are not out any money while you decide.

The feature table

Seven rows to VoiceInk, six to this one, and that is a fair reflection of what the two are. One is a larger application with a formatting layer, a dictionary and a second platform. The other is a key.

Two different bets about what dictation is

Underneath the rows there is a real disagreement, and it is worth naming because it is the thing you are actually choosing between.

VoiceInk's feature list — modes, rewriting, enhancement, phrase replacement — takes the position that raw speech is not the finished text, and that the app's job is to close the gap. Say something rambling, get something tidy. That is a defensible position and for a lot of writing it is obviously right; nobody wants their "um" transcribed.

This app takes the opposite position: what you said is the text, and anything that changes it is a thing you did not authorise and cannot see. The output is exactly the sentence, with punctuation, and the editing happens with your hands afterwards. That is worse for producing a polished paragraph from loose speech and better for anything where the words are the point — a quote, a commit message, a legal sentence, a line of prose you composed in your head and want written down unaltered.

Neither position is a feature deficiency. They lead to different products, and if you already know which one you want you already know which app to buy.

Where they overlap completely

Both do the recognition on your Mac. Neither uploads audio. Both are push-to-talk rather than always-listening. Both need the microphone and accessibility permissions macOS insists on, and both look broken until you grant them. Both are one-time purchases in a category that has largely gone to subscriptions, and that shared position is worth more than the difference between them — the honest framing of this page is two apps on the same side of the industry's biggest line, arguing about the details.

If you are here because a monthly charge annoyed you, either of these solves the actual problem. Wispr Flow at $15 a month is the comparison where that arithmetic is the whole story.

What "lifetime access" is actually promising

Both of these are sold as one-time purchases, which in 2026 is close to a political position, and both use language worth reading carefully rather than warmly.

VoiceInk says "One-time purchase, lifetime access" against each tier. Coii VoiceInput says $19 once with updates included and nothing renewing. Neither sentence is a promise that the software will be updated forever, because no company can promise that and the ones that imply it are the ones to be careful of. What both are promising is narrower and still valuable: you will not be billed again, and the copy you have keeps working.

The practical difference between a one-time purchase and a subscription is not really the money — at $19 or $49 against $180 a year it obviously is, but that is not the interesting part. It is that a subscription makes the software a tenancy. Stop paying and the key stops working, usually at the least convenient moment, and you find out how much of your working method was rented. Both apps on this page sell you the thing rather than access to the thing, and if you got here from a renewal notice, that shared property is the reason either of them solves your problem.

Where they differ is what happens if the vendor goes quiet. VoiceInk's source is public, so a determined user can build it. This app's answer is smaller: the build you have is notarised and keeps running, and updates arrive while they arrive. That is a real difference and it points at VoiceInk.

What neither of them does

The overlap in the negative column is larger than the positive one, and it saves a refund request to say so.

Neither app records or summarises a meeting. Neither transcribes an audio file you already have — for that, the larger local apps in this category are the answer and both are covered elsewhere in this library. Neither is a voice control system: you cannot say "delete that paragraph" or "switch to Safari" to either of them, and if that is what you want you are looking at a different category of software entirely. Neither offers real-time translation. Neither runs on Windows or Linux.

And neither is a substitute for a decent microphone. A built-in laptop mic in a room with other people in it will do more damage to your transcripts than any difference between these two products, and it is the cheapest fix available in this whole comparison.

The first fortnight

The pattern is consistent enough to predict, and knowing it in advance stops you abandoning a good tool in the wrong week.

The first two days are self-conscious. You will speak too carefully, over-enunciate, and produce text that reads like a voice memo. The transcripts will be fine and the sentences will be stiff.

By the end of the first week you stop performing and start talking, which is when the quality of the text goes up rather than down — dictated prose reads better when it is said the way you would say it. This is also the week the friction shows up, and it is the reason to keep a note: the times the text went to the wrong window, the times you had to wait, the times your clipboard was gone.

By the second week you are using it for the short things, and that is the actual switch. Nobody's writing changes because they dictated a long document once. It changes when the two-line replies, the commit messages and the search-box sentences stop being typed. Sixty presses a day is what a converted user looks like, and no evaluation shorter than a fortnight will show you whether you get there.

Which is why a thirty-day trial and a fourteen-day refund window are not quite the same offer. Both are generous. Only one of them lets you find out slowly without money having changed hands.

How to test them against each other

Install one. Set its hotkey. Use it for everything for a week, including the things you would normally type — the two-word replies, the commit messages, the search boxes. Then swap.

Change one of the hotkeys before you run both, because two dictation apps listening for the same key is the one configuration that breaks reliably and will look like a fault in whichever you happen to be blaming.

Keep a list of every time the text went somewhere you did not expect, arrived noticeably late, or did not arrive. That list is the comparison. Transcript quality will be fine in both and will tell you nothing, and the failures are what you will still be living with in six months.

One more thing worth testing deliberately rather than waiting to discover: pick the three applications you write in most and dictate a full paragraph into each. Some applications accept inserted text cleanly, some are fussy about it, and some do something strange with the cursor afterwards. Which ones those are varies by machine and by version, and it is much better to learn it in week one of a trial than in month three of ownership.

What switching costs

Nothing structural, in either direction. Neither app owns a library, a document format or an archive you would be trapped inside; the output has always been plain text and it went straight into your own files.

Moving here means an install, two permissions, one hotkey, and about ten minutes. Moving away means dragging the app to the bin. The only thing worth carrying is anything you want to keep out of the history list, which is a local list and copies out by hand.

The real cost is the custom dictionary, and it only runs one way. If you have spent months teaching VoiceInk your vocabulary, that work does not come across and there is nothing here to receive it. Weigh that honestly: it is the single strongest reason not to switch, and it gets stronger the longer you have been using it.

Which one to buy

VoiceInk if you have an Apple Silicon Mac on macOS 14.4 or later and you want the dictionary, the modes, the rewriting or the iPhone app — or if "built in public on GitHub" is the sentence you have been looking for. It is more app for the money and it is honest about where its optional cloud step begins.

This one if your Mac is Intel or on Ventura, you dictate across three machines, you want the words written down before anything can go wrong with them, or you want the output to be exactly what you said with nothing shaping it. Thirty days of the whole app, no card and no account, is the only test that settles it.

The neighbouring pages, if this pair is not your pair: Murmur is the free open-source option, BetterDictation is the other one-time purchase in this bracket, and superwhisper and MacWhisper are the larger local apps that transcribe files as well.