Aqua Voice vs MacWhisper on Mac in 2026: dictation, privacy and prices compared

What Aqua Voice and MacWhisper are built to do

Aqua Voice is built around dictation. You start it with a keypress, speak, and the text lands at your cursor. The company's homepage lists paste at the cursor, live refinement, a custom dictionary, style rules and 49 languages. Its Edit Mode and Deep Context features aim at the same job, which is making the text fit the place you type it.

MacWhisper is built around transcription. Its center is audio and video files, YouTube links, podcasts and meetings. The official site also lists system-wide dictation, 100+ languages with translation, and AI summaries and chat. Those AI features run on local models such as Ollama or LM Studio, or on your own API keys.

That split shapes almost every row in the comparison below. If most of your day goes to typing messages and documents, Aqua Voice is the closer fit. If most of your day goes to turning recordings into text, MacWhisper is.

Price: what each app costs to start

Aqua Voice has a free plan with 1,000 words in total. Pro removes that word cap, and the pricing page calls it unlimited words. Pro also adds Custom Instructions and a Personal Dictionary. The higher tiers, Max and Business, add features such as Voice Commands and team billing. Paid prices change, so check the live pricing page before you buy.

MacWhisper's free tier is listed as free forever. Pro is a one-time €64 payment, and it includes lifetime updates. You can buy it from the Gumroad page or the Mac App Store. A TheSweetBits reviewer calls the free version genuinely useful. The same reviewer says cloud and AI services can add separate costs.

Is MacWhisper free, and when does Aqua Voice's free plan run out?

Free plans look similar until you use them. Aqua's free plan counts words in total, not per day. Once you use the 1,000 words, you need Pro to keep dictating. A normal week of work can use that up faster than you expect, so note your count from the first day.

MacWhisper's free tier is the cheaper way to test the transcription side. Run one of your own recordings through it before you pay €64. That single test tells you more than any demo video.

• Write down how many words you dictate in a normal week.
• Check whether you need Pro for Custom Instructions or a Personal Dictionary.
• For MacWhisper, decide between the direct download and the App Store edition before you pay.
• Add up cloud or AI service costs separately, since they are not in the base price.

Privacy: where your audio and text go

Aqua Voice's privacy policy turns on one setting, Privacy Mode. With Privacy Mode off, Aqua says it securely stores dictation audio and transcripts on its servers. With Privacy Mode on, it says it does not store them on its servers. The homepage puts it in one line: turn on Privacy Mode and nothing is stored on its servers at all.

Two details deserve a close read. The policy says Notetaker stores meeting audio, transcripts, notes and screenshots even when Privacy Mode is on. The policy also does not say whether data trains models, what encryption it uses, or where processing happens. Aqua's pricing page says it holds a SOC 2 Type II audit report and ISO 27001 certification.

MacWhisper's privacy policy is shorter. It says all functions run on your device, and no audio, text or other data leaves it. Its analytics cover only anonymous licensing events. The policy shows a last-updated date of February 14, 2024, which is before the cloud features arrived.

The support docs tell a more detailed story. Transcription, including speaker identification, runs locally by default. Cloud providers such as Groq, ElevenLabs and Deepgram send audio off your Mac. So do DeepL translation and AI prompts sent to services like OpenAI or Anthropic. Local Ollama or LM Studio keeps your data on the Mac.

Offline use and the hardware you need

MacWhisper is the clear pick for offline work. Local transcription is the default, so the work stays on your Mac. The Medium and Large models need more than 8GB of RAM, according to its dictation page. That page also warns that older Intel Macs may run slower.

Aqua Voice needs a connection. Reviewer SoftOrbits says every phrase is processed on Aqua's servers, and that Aqua has no on-device mode at any tier. That is one reviewer's reading, not a company statement. Test it yourself. Turn Wi-Fi off from the menu bar icon, then dictate a sentence.

Hardware matters on the iPhone side too. MacWhisper's iOS docs say on-device models need iPhone 15 or newer, and the team plans to lower that to iPhone 13. Cloud transcription through Groq has a 25MB file size limit.

Dictation: how each app types into any app

Reviewers rate Aqua Voice highly for dictation. Aestumanda tested it on a Mac in August 2025. The reviewer called it 'the most accurate and productive dictation tool I have ever used.' The same review says it surpasses Dragon Professional. It also praises near real-time transcription and the custom dictionary.

The same review lists complaints. Starting dictation and pasting still need keypresses. The floating box 'can feel discordant.' In noisy shared spaces, Aqua 'will often transcribe those voices alongside mine.' Those points matter if you work in an open office.

MacWhisper dictates system-wide, but only in the direct download. The Mac App Store edition does not include that feature, according to TheSweetBits. The reviewer writes that dictation 'has become our favorite way to use MacWhisper.' Privacy is the reason they give for that, along with speed.

Proper nouns and brand names are the weak spot for MacWhisper, according to the same review. If you dictate client names or product names all day, that matters more than any feature list. Test those names first, before you commit to either app.

Languages and mixed-language dictation

Aqua lists 49 languages on its homepage. MacWhisper lists 100+ languages, with translation. If you switch languages during the day, check each app's language list before you start. Translation in MacWhisper can go through DeepL, which sends text off your Mac, so keep it off for sensitive text.

Accuracy on names, jargon and code

Treat both accuracy claims with care. Aqua's homepage says its Avalon model scores 97.4% key-term accuracy on its own AISpeak-10 benchmark. That is a vendor claim, not an independent test. SoftOrbits cites 97.3% for the same benchmark and makes the same point.

SoftOrbits also says Aqua is 'quick and accurate on code and technical terms.' TheSweetBits reports weaker results for MacWhisper on proper nouns. Neither set of results comes from a public test that you can repeat. That is why your own short test beats both the reviews and the vendor pages.

A short accuracy test you can run today

1. Write a short list of sentences that use your real names, product words and code terms.
2. Dictate each sentence in Aqua Voice into TextEdit, with Format > Make Plain Text turned on. Count the sentences that need a fix.
3. Repeat the same list with MacWhisper's dictation, using the same microphone and the same room.
4. Compare the sentences that failed, not the total word count.
5. Note which app kept your names spelled right without help.

If Aqua Voice misses a name, add that name to the custom dictionary. On Pro, that is the Personal Dictionary. Then run the list again. If MacWhisper misses a name, the fix is more work, so decide whether that cost is worth it for your job.

Files, podcasts and meeting recordings

MacWhisper is built for recordings. It transcribes audio and video files, YouTube links and podcasts. It adds speaker recognition, subtitle export and batch processing. Watched folders and a CLI let you automate the work. For most recording jobs, that is the main reason to pick it.

MacWhisper also records meetings on Zoom, Teams and Webex without a bot, according to its homepage. Aqua Voice has Notetaker instead. Its policy says Notetaker stores meeting audio, transcripts, notes and screenshots, even with Privacy Mode on.

The Aqua pages I checked describe dictation and Notetaker, not file transcription. Check Aqua's site before you rely on Aqua Voice for recordings. If a file is sensitive, keep the storage rules in mind for both apps.

Worked example: a weekly podcast

Say you record a podcast each week. Each episode needs a transcript and subtitles. Open MacWhisper and drop in the recording. Turn on speaker recognition, and keep the transcription engine on a local model. Export the subtitles when the job finishes.

Aqua Voice is built for dictation, so it is a weaker fit for this job. Its pages describe dictation and meeting notes, not a file import workflow. Pick MacWhisper here, and keep the cloud and translation switches off unless you accept what they send.

Platforms: Mac, Windows, iPhone and Android

Aqua Voice runs on Mac, Windows, iOS and Android. Its changelog lists the Android app at version 0.1.0 on October 6, 2026, which needs Android 10 or later. The Mac and Windows apps reached version 0.21.4 on October 17, 2026.

MacWhisper covers the Mac and the iPhone. The App Store listing needs macOS 15.0 or later. Its pages I checked do not list a Windows or Linux build, so do not plan your work around one.

Lazur runs on the Mac only, on Apple silicon and Intel. Windows is planned, not available. If your team works on Windows, Aqua Voice is the only app in this comparison that covers it.

Who should pick Aqua Voice

Pick Aqua Voice if you dictate into many apps across several devices. It also fits you if you can accept cloud processing and want a custom dictionary, style rules and 49 languages. Turn on Privacy Mode before you start, and check the Notetaker settings separately, since they work differently.

Who should pick MacWhisper

• You transcribe recordings, podcasts, lectures or meetings on a Mac.
• Your audio must stay on the device, and you keep the cloud switches off.
• You want a one-time €64 purchase with lifetime updates.
• You have more than 8GB of RAM and want the Medium or Large models.

Common mistakes when choosing between them

• Assuming Privacy Mode covers everything. Aqua's policy says Notetaker stores data even when Privacy Mode is on.
• Buying MacWhisper's App Store edition for system-wide dictation. Only the direct download includes it.
• Sending sensitive recordings to a cloud provider without checking the setting first.
• Treating 97.4% as a neutral score. It is Aqua's own benchmark, not an independent result.
• Forgetting microphone access. Check System Settings > Privacy & Security > Microphone before you test either app.
• Copying old prices from reviews. SoftOrbits and Aestumanda cite Aqua Pro at $8 per month, which conflicts with the pricing page.

Most of these mistakes come from reading one page and stopping. Read the privacy policy, the support docs and one independent review before you decide. Then run your own test with your own names and files.

Where Lazur fits if you want voice help

Lazur is a third option, and it works differently from both apps. It is a voice assistant for the Mac that lives in the notch. Hold Option + Command (⌥⌘) to ask about anything on screen. Lazur answers out loud. Ask where something is, and it draws a circle and a label around it.

Dictation is included, but it is not the headline. Hold fn to dictate clean, formatted text at the cursor in any app. Snippets, a personal dictionary and clipboard slots come with it. Lazur also keeps a memory of what you asked and did, kept on the Mac.

Lazur does not click buttons, edit files or operate other apps for you. When you ask how to do something, it guides you one click at a time. Its facts do not include file transcription, so pick MacWhisper for recordings. For a side-by-side look, read Lazur vs MacWhisper.

Next steps before you choose

1. Open System Settings > Privacy & Security > Microphone and allow each app you plan to test.
2. In Aqua Voice, turn on Privacy Mode and dictate one sentence that includes a name.
3. In MacWhisper, download the direct edition from its site and run one of your own recordings on a local model.
4. Run the accuracy test with the same microphone and the same room.
5. Read the privacy setting for any cloud or AI service before you record client or patient audio.