Aqua Voice Review (2026): Pricing, Limits and What It Actually Does
Aqua Voice is hold-a-key dictation for the Mac that cleans up your speech as you talk, aimed squarely at people who write prompts all day.
Open Supavocal → Official site
The core idea, minus the marketing
Hold the space bar, talk, let go. Text appears wherever your cursor was. The framing on the site is that we have typed for a hundred and fifty years and it is time to speak, which is a nice line, but the actual product decision underneath it is smaller and smarter: dictation as a system-wide modifier key rather than as an app you switch into. That is why the homepage shows it firing inside Gmail, Google Docs and Slack instead of inside its own window. The moment dictation requires a context switch, people stop using it. Making it a held key removes the switch, and most of the product's value comes from that one choice.
Speed claims, and how to read them
Two numbers get top billing. The company says five times faster than typing and twice as accurate, and the on-page demo compares Aqua 3.1 at 230 words per minute against a keyboard at 40. Those are vendor-run figures under vendor-chosen conditions, so treat the ratio as directional rather than as a benchmark. The more interesting half of the demo is what it shows going wrong on the keyboard side: a repeated and misspelled hook name in a React snippet. That is the real pitch. Dictation is not just faster, it skips the specific errors you make when typing technical words quickly, provided the model recognises them in the first place.
Why it aims at people writing prompts
A large part of the site is given over to coding and prompting rather than to email. The claim is that the system understands syntax, libraries and framework names as you speak, and there is a feature described as your screen being its dictionary, meaning it reads on-screen context to resolve words it would otherwise mangle. Logos for the Codex app, Cursor and Claude sit under a line about giving a coding model full context by speaking your prompt. Whether that context awareness holds for your own codebase and your own variable names is exactly the thing to test in the first ten minutes, because it is the difference between a novelty and a tool.
Platform, price and the parts not on the page
Every call to action on the homepage is get started on Mac. There is a platforms menu and a pricing page in the navigation, but the crawl shows no plan table, so check the official pricing page for current rates rather than trusting a number from a review. Company logos for Apple, Amazon, Notion and Perplexity appear under a trusted-by heading, which tells you the vendor has recognisable users and nothing about what those companies actually deploy. If you are on Windows or Linux, confirm availability on the platforms page before you get attached to the idea. The same question decides any team rollout, because a Mac-only tool quietly splits your team into the people who get the speed increase and the people still typing at forty words a minute.
The parts that matter day to day
Hold a key, speak, done
Dictation triggered by holding the space bar rather than by opening an app, so it works wherever the cursor already is and does not interrupt whatever you were doing.
Cleanup while you talk
The text is refined as you speak rather than dumped raw and edited afterwards, which is the difference between usable output and a transcript you have to rewrite anyway.
Context from your screen
The product claims to read what is on screen to resolve code syntax and unusual terms, which is the feature most likely to decide whether it works for technical dictation.
Built for prompting
Positioned around coding assistants and prompt writing, with the pitch that speaking gives a model more context than you would bother typing out by hand.
Different directions: Aqua Voice and Supavocal
| Feature | Aqua Voice | Supavocal |
|---|---|---|
| Direction | Speech into text | Text into speech |
| Voice cloning | Not part of the product | Yes, a core feature |
| Where it runs | Mac, per the download on the site | A studio you open in the browser |
| Typical use | Dictating prompts, emails and documents | Narration, voiceover and character voices |
| Works inside other apps | Yes, wherever your cursor is | Not applicable, output is an audio file |
| Pricing | See the official pricing page for current rates | See supavocal.com for current plans |
A ten-minute test that tells you the answer
- Dictate a prompt with real names
Speak a prompt containing your own function names, your framework and a library nobody outside your team uses. Generic dictation handles plain English fine, so test the words that usually break it. - Try it in three different apps
Run it in your editor, your mail client and your chat tool. A system-wide feature is only system-wide until it meets the one app that steals the key binding you rely on. - Compare against your real typing
Time yourself writing the same paragraph by hand. Vendor benchmarks compare against an average typist, and you may already be faster than the number the comparison assumes. - Check the pricing page directly
No plan table appears on the homepage, so open the pricing page before the trial ends and confirm what a month costs on the platform you actually use.
Questions worth asking first
Does Aqua Voice work on Windows?
Every download button on the homepage points at the Mac version, and there is a separate platforms menu in the navigation. Check that page for current availability rather than assuming parity. If you are on Windows or Linux and the answer is no today, the rest of the evaluation is academic regardless of how good the dictation is.
How much does Aqua Voice cost?
A pricing page exists but the homepage carries no plan table, so any number quoted in a review may be stale. Open the official pricing page and check current rates before committing. If a trial is offered, use it on the apps you genuinely work in rather than on a test document that proves nothing.
Is it really five times faster than typing?
That figure and the 230 words per minute demo come from the vendor under conditions it chose. Speaking is faster than typing for most people, so the direction is believable. Whether the multiple holds for you depends on your typing speed and on how much editing the transcribed text still needs afterwards.
Does it handle code and technical terms?
The product is marketed on exactly that, with a claim that it understands syntax, libraries and frameworks, and a feature that reads on-screen context for unfamiliar words. It is also the first thing to test, because recognition of your own internal naming is where every dictation tool either works or quietly does not.
Can Aqua Voice generate a voice for my video?
No, and it does not claim to. It converts speech to text, which is the opposite direction. For narration, voiceover or a cloned voice you can reuse across videos, Supavocal at supavocal.com is built for that side of the problem.
Does it work offline?
The site does not state an offline mode either way, so assume a connection is needed until the documentation says otherwise. This matters more than it sounds if you dictate on planes or in buildings with poor coverage, and it is a fair question to send to support before paying for a year.
Need a voice, not a transcript?
Supavocal turns written text into natural speech and can clone a voice you reuse across projects. If dictation solved the wrong half of your problem, start here.
Open Supavocal →