What Mac dictation apps shipped lately, by date

By · Published · Updated

Three changes in the last eight weeks matter more than the rest, as of October 6, 2026: Aqua Voice released an Android app on October 6, Wispr Flow moved every user to Canto, a speech model it built itself, on September 17, and superwhisper made all of its local Whisper models free on the Mac on August 26. The rest of this page is the dated list behind that sentence, app by app.

I built YapToText, a free dictation app that runs on the Mac, so its own releases are on the list too, and you should read the page knowing I made one of the apps on it. My hands make typing difficult, so I dictate every day and I follow these changelogs anyway. Every entry comes from the maker's own changelog, release notes or site, read on October 6, 2026, with the date the maker gives. I have not tested the other apps hands-on, so nothing here says whether a change works well.

Download YapToText on the Mac App StoreThis app is completely free. You don't have to make an account. There are no subscriptions, no difficult downloads, and nothing is locked. Made for Apple silicon Macs. Source on GitHub.

On this page: 10 sections
  1. The changes that matter most
  2. Wispr Flow updates
  3. superwhisper updates
  4. Aqua Voice updates
  5. VoiceInk, Monologue, Spokenly, Typeless and MacWhisper updates
    1. VoiceInk
    2. Monologue
    3. Spokenly
    4. Typeless
    5. MacWhisper
  6. Dictation built into Raycast and VS Code
    1. Raycast
    2. VS Code
  7. YapToText updates
  8. Prices and platforms that stayed the same
  9. How this log is kept
  10. Questions

The changes that matter most

Read October 6, 2026, on each maker's own changelog. Dates are the makers' dates.

DateAppWhat changed
October 6Aqua VoiceAn Android app on Google Play, for Android 10 or later, covered by the same subscription
September 29YapToText1.5.3 on the Mac App Store: the dictation and Quick Edit pop-ups can be dragged again on macOS 27
September 21TypelessHelp me write drafts a reply, an email or a post from spoken instructions
September 17Wispr FlowFlow runs on Canto, the first speech model Wispr built itself, for every user
September 15Wispr FlowNotetaker, its meeting notes product, arrives on Windows
August 26superwhisperAll local Whisper models are free on macOS, and the Pro trial becomes 3,000 words instead of 15 minutes
August 25RaycastDictation, free during the Raycast 2 beta, becomes part of Raycast Pro
August 19superwhisperS1-mini, a language model that cleans up dictation on the device
August 12VoiceInkRefine, a cleanup model that runs on the Mac
July 29VS CodeBuilt-in dictation, experimental, with a speech model that runs on the device
July 20Aqua VoiceEdit Mode: select text, hold the Aqua key and say the change

The direction is the same in three of these apps: more of the work is moving onto the device, or into a model the maker owns. The full comparison has what each app costs and where your voice goes today.

One change has no line in the table, because Apple gives it no dated changelog: in macOS 27 Golden Gate, Apple lists improved accuracy in Dictation from a new on-device model, on Macs with M3 or later and at least 12 GB of memory, in English. The guide to Dictation in macOS 27 has which Macs get it.

Wispr Flow updates

Wispr Flow publishes a What's new page with dated posts and version notes. Its biggest change this season is under the hood: the model that hears you is now its own.

  • September 17. Flow now runs on Canto, which Wispr calls the first speech model it built itself, rolled out for everyone with nothing to switch on. Wispr says Canto had the lowest word error rate of the models it tested on its own recordings of real Flow use, in English. That is Wispr's test, and I have not repeated it. Transcription still happens in the cloud.
  • September 15. Notetaker is available on Windows as well as the Mac.
  • September 4. Version 1.6.721 imports Otter transcripts into Notetaker, and the desktop app now holds a pending update until you step away instead of restarting mid-task.
  • August 28. Version 1.6.675 adds an org-wide switch for Notetaker and Growth plans that a team can buy without a sales call.
  • August 21. Version 1.6.580 lets you pick virtual and routing audio devices, such as Krisp or BlackHole, as the dictation microphone.
  • August 5. Notetaker launches, which Wispr calls its first product beyond dictation.

Prices did not move: Pro was still $15 per user a month, or $12 billed yearly, on October 6. The Wispr Flow pricing page has every plan, and the Wispr Flow alternative page covers the switch.

superwhisper updates

superwhisper keeps a changelog by platform. These entries are from the macOS tab. Its free plan got bigger in August, and its cleanup started to move onto the device.

  • October 5. 2.19.1 and 2.19.2 fix a crash on a fresh install and updates that could get stuck while installing.
  • October 2. 2.19.0 starts the microphone faster, adds a Muffle option that filters other apps' audio while you record, and lowers playback to 30 percent of your current volume instead of a fixed level.
  • September 3. 2.18.3 turns silence removal on by default, keeps your clipboard contents after a dictation by default, and fixes several problems on macOS 27.
  • August 26. 2.18.2 makes all local Whisper models free on macOS, and changes the free Pro trial from 15 minutes to 3,000 words.
  • August 19. 2.18.0 adds S1-mini, an on-device language model that cleans up and formats dictation without a network request. Its documentation says S1-mini is English only, works in the Message and Mail modes, and needs Pro.
  • August 6. 2.17.2 introduces the S1 family of models.

Prices did not move: Pro was still $8.49 a month, $84.99 a year or $249.99 once on October 6. One thing to check before you install on an older Mac: current releases need macOS 14 or later, and its troubleshooting guide names 2.16.1 as the last release for macOS 13.3, though its download page still says 13.3. The superwhisper pricing page has the plans, and the superwhisper alternative page has the comparison.

Aqua Voice updates

Aqua Voice keeps a changelog and a machine-readable summary of what is new. With Android, Aqua now runs on the same four platforms as Wispr Flow.

  • October 6. Aqua Voice for Android reaches Google Play, for Android 10 or later. A floating button sits next to the keyboard you already use, and Aqua says an existing subscription covers it. Realtime Mode and Send It stay on the desktop.
  • September 23. Aqua says it is GDPR and HIPAA compliant, next to its SOC 2 Type II and ISO 27001 certifications. Those are Aqua's statements. Ask it about terms before you use it with health information.
  • September 22. Mac version 0.20.0 adds a translucent floating bar on macOS 26 and later, and iPhone version 1.0.37 fixes crashes.
  • September 14. Windows version 0.19.14 starts the microphone sooner. Aqua gives 122 ms, down from 298 ms in late August, by its own measurement.
  • September 4. Version 0.19.12 lets you block microphones, such as virtual inputs, so Aqua picks another one.
  • August 27. The iPhone app can start dictation from the Action Button.
  • July 26 and July 30. Send It submits a message when you end a Realtime dictation with those words, on the Max plan, and a mouse button can start Aqua.
  • July 20. Edit Mode arrives in version 0.17.0, for every desktop user.

Prices did not move: Pro was still $8 a month billed yearly, or $10 month to month, on October 6. The Aqua Voice pricing page has every plan, and the Aqua Voice alternative page has the comparison.

VoiceInk, Monologue, Spokenly, Typeless and MacWhisper updates

Five more apps shipped dated releases in the same weeks. VoiceInk, Monologue, Spokenly and Typeless ship every week or two, and MacWhisper's latest release is about which AI services it can call.

VoiceInk

  • October 1. 2.22 is a diagnostic prerelease, published to investigate Parakeet transcription failures and hangs.
  • September 28. 2.21 adds the Parakeet Ultra model, OpenRouter transcription models, double-tap recording shortcuts, and Auto Send as a global setting.
  • September 19. 2.20 adds Dictionary Auto Learn, dictionary import and export, a Quick History of recent transcriptions, and mouse shortcuts.
  • August 12. 2.11 adds VoiceInk Refine, for cleanup on the device, and experimental support for Cohere Transcribe.

Monologue

  • October 2. iPhone version 1.2 finds and opens notes through Siri AI, Spotlight and Shortcuts on iOS 27.
  • September 28. Mac version 1.5.1 finishes and pastes a dictation faster, adds Compose in beta for writing or editing by voice, and offers 4-hour and 6-hour limits for recorded notes.

Spokenly

  • October 5. 2.30.7 adds Phonon-2, a local model that Spokenly describes as fast, compact and English only, and 2.30.8 adds a compact pill-shaped recording display with live text.
  • October 2. 2.30.6 makes the app available in Italian and Brazilian Portuguese, and lets an organization enforce Local-only mode with an MDM profile.
  • September 27. 2.30.2 adds the Parakeet Ultra and Parakeet Redux local models, and meeting detection in Safari.
  • September 12. 2.29.3 can start dictation when you tap your MacBook, and adds a shortcut for meeting recordings.

Typeless

  • September 21. Mac version 2.8.0 adds Help me write: press the Ask anything shortcut and say what you need, and Typeless drafts a reply, an email, a post or a document from your instructions.
  • September 15. 2.7.0 lets a mouse side button or middle click start Dictate, Translate or Ask anything.
  • August 25. 2.4.0 adds Cloud Sync, which keeps your dictation history across devices for as long as you choose.
  • August 10. 2.3.0 adds activity insights: active days, streaks and a usage heatmap.

MacWhisper

  • September 29. Version 5.2 of Whisper Transcription, its App Store edition, adds newer AI models and one more AI service, and raises ElevenLabs cloud transcription to files of up to 3 GB and 10 hours.

Prices did not move on October 6: VoiceInk was still $25 once for one Mac, Monologue $15 a month or $144 a year, and MacWhisper Pro 64 euros once. The VoiceInk, Monologue and MacWhisper alternative pages have the comparisons.

Dictation built into Raycast and VS Code

Two tools that are not dictation apps now have dictation inside them. Raycast charges for its dictation. VS Code's runs on the device and is still marked experimental.

Raycast

  • October 7. 2.7: Dictation shows a clearer status when a session wraps up, and Raycast says its post-processing is faster and more reliable.
  • September 28. 2.5: dictation with Control M in AI Chat and Quick AI uses the same processed transcript as the Dictate command, so the text appears once you stop recording.
  • August 25. 2.0: Raycast 2 leaves beta. Dictation and AI Chat, free during the beta, become part of Raycast Pro, with a 7-day Pro trial on sign-in.
  • August 19. 0.71: Dictation can import your vocabulary and hotkeys from other dictation apps, and gains an Add Word to Vocabulary command.

VS Code

  • September 9. 1.137 adds Voice Mode, experimental: a spoken conversation with an agent, for eligible individual GitHub Copilot plans.
  • September 2. 1.136 lets administrators require on-device transcription and turn off language-model cleanup of transcripts by policy.
  • August 5. 1.132 makes the multilingual Nemotron 3.5 the default on-device dictation model, and terminal dictation keeps shell syntax.
  • July 29. 1.131 adds built-in dictation, experimental, in chat, editors and the terminal, without the Speech extension.

The Raycast Dictation page has its price and where the audio goes, and the VS Code dictation guide has the shortcuts and what its cleanup step sends.

YapToText updates

YapToText is my app, so this is the one list I can vouch for line by line. Every release is free, and both models ship inside the App Store app.

  • September 29. 1.5.3 (21), on the Mac App Store: the dictation pop-up and the Quick Edit pop-up can be dragged again on macOS 27, and media pausing was rebuilt, so a paused video stays paused when you dictate. GitHub and Homebrew are still on 1.5.2 (19).
  • September 18. 1.5.2 (19): language auto-detection, Polish and fourteen more languages, a warning when the room is too loud for the microphone, and Bluetooth hearing aids and headsets that keep their full sound quality. Apple Speech was removed, so Whisper is the only speech engine.
  • September 10. 1.5.1 (16): dictating into a terminal works better, and a canceled dictation is still transcribed and saved to History instead of thrown away.
App Store poster headed Nothing leaves your Mac, over YapToText's AI Models page with both models marked Included.
Figure 1. In 1.5.3, both models ship inside the App Store app and show as Included.

The release notes go back to 1.0, in the app's own words.

Prices and platforms that stayed the same

A changelog only says what moved. These are the things I reread on October 6 that had not changed.

  • Typeless. Free to 8,000 words a week, and Pro at $12 per member a month billed yearly or $30 month to month. Its FAQ still says every feature needs an internet connection.
  • Willow Voice. Pro at $15 per user a month, or $12 billed yearly, with Android still listed as coming soon.
  • Spokenly. Local models free, Pro at $99.99 a year, and an Android page that says it is not yet available, with a waitlist.
  • Monologue. Windows and Android still on a waitlist.

If you are choosing between these apps today, the roundup of AI dictation apps ranks ten of them with the same facts.

How this log is kept

This page is only useful if it is right, so here is how it is kept. Every line is the maker's own words on the maker's own date, and I reread each changelog once a week.

  • Where entries come from. The changelog, release notes or What's new page each maker publishes, listed under Sources with the day I read it. VoiceInk's entries are from its GitHub releases, and MacWhisper's from the App Store listing of Whisper Transcription.
  • What is left out. Small fixes, and anything I could not read on the maker's own pages. For Typeless I read the Mac release notes only, not the Windows, Linux or phone ones.
  • What I have not done. Tested any of these changes. A line here means the maker says it shipped, not that I measured it.
  • Corrections. If a line is wrong or out of date, tell me on GitHub. Every issue gets read.

Download YapToText on the Mac App StoreThis app is completely free. You don't have to make an account. There are no subscriptions, no difficult downloads, and nothing is locked. Made for Apple silicon Macs. Source on GitHub.

About the author. Ryleigh Newman is a developer whose hands make typing difficult and who dictates every day with YapToText, the app this site describes. Facts about other apps come from their own sites, with the date they were read. More about the project.

Sources

Questions

Does Aqua Voice have an Android app?

Yes, since October 6, 2026. Aqua says Aqua Voice for Android is on Google Play for Android 10 or later, and that one subscription covers Mac, Windows, iPhone and Android.

What speech model does Wispr Flow use?

Canto, a speech model Wispr built itself. Its What's new page says Flow has run on Canto since September 17, 2026, for every user. Transcription still happens in the cloud.

Are superwhisper's Whisper models free?

Yes, on the Mac. superwhisper's changelog says version 2.18.2, released August 26, 2026, made all local Whisper models free on macOS. Its AI modes, file transcription and cloud models are still Pro.

What is superwhisper S1-mini?

An on-device language model that superwhisper added in version 2.18.0 on August 19, 2026, to clean up and format dictation without a network request. Its documentation says S1-mini is English only, works in the Message and Mail modes, and needs Pro.

When was MacWhisper last updated?

Version 5.2 of Whisper Transcription, MacWhisper's App Store edition, is dated September 29, 2026, on its App Store listing, read on October 6, 2026.

What is the latest version of YapToText?

1.5.3 (21), on the Mac App Store since September 29, 2026. GitHub and Homebrew are still on 1.5.2 (19).