Skip to content
Petal

AlternativesMac6 min read

Aqua Voice alternative that works offline

Need an Aqua Voice alternative that works offline? Petal for Mac is free, runs speech on your Mac, and sends text only to a provider that you pick.

Published by Optimal Apps, the maker of Petal.

On this page

The best Aqua Voice alternative for offline use is a dictation app that runs the speech model on your Mac. Petal for Mac is free and open source. It transcribes on the Mac with downloaded models, cleans up the text on the Mac with Petal W1 if you want, and pastes the result into the front app. There is no account and no word cap. If you add a cloud model, Petal sends the transcript only to the provider that you pick, with your own key.

Aqua Voice is a fast, polished app, and its paid plans have real features. This guide compares the two fairly and shows how to set up Petal on macOS 26.

Quick checklist: Aqua Voice vs Petal

Question Aqua Voice Petal for Mac
Works with no internet Not listed; described as cloud only Yes, after the model download
Free plan 2,000 words Free, no word cap
Paid plans Pro $10–12/month, Max $19–24/month, Business $19–24/seat/month None on Mac
Account Required Not required
Zero data retention Business plan Speech stays on the Mac by default
Platforms Mac, Windows, iOS keyboard, Android Mac; separate iPhone and Windows apps
License Closed source MIT, open source

Prices are from the Aqua Voice pricing page at the time of writing. Check aquavoice.com/pricing for current numbers.

If you want to keep Aqua Voice and only need a fix, read Aqua Voice not working first.

What Aqua Voice does well

Aqua Voice has a fast cloud model (Avalon), context features, a personal dictionary on Pro, and a Realtime Mode and voice commands on Max. Its Business plan adds SSO, shared dictionaries, and zero data retention. Aqua Voice says that it has a SOC 2 Type II report and ISO 27001 certification. For a team that is fine with a cloud service, these are good reasons to stay.

Why people look for an Aqua Voice alternative

It needs the network

Aqua Voice sends audio to its servers to transcribe. Its own guide says that it needs HTTPS and WebSockets on port 443, so some corporate proxies and VPNs break it. On a plane, on a train, or on a weak connection, dictation stops.

The free tier ends fast

At a normal speaking rate of 130 to 150 words per minute, 2,000 words is about 13 to 15 minutes of speech. After that, you need a paid plan.

Privacy on personal plans

On the pricing page, zero data retention appears only on the Business plan. Third-party reviews say Privacy Mode is off by default for individual users. If you dictate private notes, client data, or code, you may prefer that the audio does not leave your Mac.

Terminal and editor problems

Users report that dictated text sometimes does not appear in some terminals, such as Claude Code (anthropics/claude-code issue #51725). Aqua Voice suggests "Paste Last Transcript" in those cases. For more about terminals, read dictation not working in VS Code and Terminal.

How Petal works as an offline Aqua Voice alternative

Petal for Mac is a menu bar app from Optimal Apps, the publisher of this site. The source code is on GitHub at Aayush9029/petal under the MIT license, so anyone can check what it sends.

Speech runs on the Mac

You download a speech model once. Then Petal transcribes on the Mac with FluidAudio (Core ML), WhisperKit, or MLX:

Model Size Best for
Parakeet 110M about 455 MB Fast English dictation
Parakeet 0.6B V3 about 485 MB 25 European languages
Whisper Large V3 Turbo about 1.1 GB 99 languages
Qwen3 ASR 1.7B about 2.5 GB High accuracy in 30 languages

Cleanup on the Mac, or with your own key

In Settings > Intelligence, you choose how Petal cleans up the text:

  • Petal W1 runs on the Mac. It removes fillers, repeats, and false starts and keeps your voice. The Petal README says it does this in about 0.2 seconds.
  • Apple Intelligence runs on Macs that support it.
  • Cloud Model uses OpenAI, Anthropic, or OpenRouter with your own key, or a custom server such as Ollama or LM Studio. A local Ollama or LM Studio server keeps this step on your Mac too.
  • Off pastes your words as you said them.

Petal keeps API keys in the macOS keychain. With a cloud model, Petal sends the transcript, not the audio, to that provider. If you turn on the Screen tool, Petal also sends a screenshot. Screen is off unless you turn it on.

Per-app routes

In Settings > Router, each app or website can have its own route. For example, use Petal W1 in Slack, a cloud email prompt in Mail, and paste as said in Terminal. Read per-app dictation cleanup on Mac.

Paste that waits for slow apps

Petal pastes with a simulated Command-V and then restores your old clipboard after 700 ms, because Electron apps such as VS Code read the clipboard late. You can turn off Restore Clipboard in Settings > Recording. If a paste fails, the Paste Last Transcription shortcut pastes the last result again.

How to switch from Aqua Voice to Petal

  1. Quit Aqua Voice, so the two apps do not fight over the same shortcut.
  2. Download Petal from GitHub Releases. Petal needs macOS 26.2 or later on Apple silicon.
  3. Move Petal to the Applications folder and open it.
  4. Allow the microphone when macOS asks.
  5. Open System Settings > Privacy & Security > Accessibility and turn on Petal.
  6. In Petal Settings > Models, download Parakeet 110M for English, or Parakeet 0.6B V3 for other languages.
  7. In Settings > Intelligence, select Petal W1 and let it download.
  8. In Settings > General, set the Recording Shortcut. The default is Option-Space. If you liked a single-key trigger in Aqua Voice, switch the shortcut to the double-tap mode and pick a key.
  9. Turn off Wi-Fi, open a text field, press the shortcut, speak, and stop. The text appears with no network.

Aqua Voice privacy alternative: what leaves your Mac

Step Aqua Voice Petal, default Petal, Cloud Model on
Audio Sent to Aqua servers Stays on the Mac Stays on the Mac
Transcript Processed in the cloud Stays on the Mac Sent to your chosen provider
Screen context Depends on Aqua's Deep Context settings Not taken Sent only if Screen is on
Account data Aqua account None Your provider account

This table describes what each app is designed to send. Petal's code is public, so you can check it.

Limits of Petal compared with Aqua Voice

  • No voice commands. Aqua Max has voice commands. Petal does not.
  • No iPhone keyboard in the Mac app. Petal - AI Voice Recorder is a separate iPhone app with its own keyboard. See Petal for iPhone.
  • Model download. The first setup needs a download of a few hundred MB. Aqua Voice needs no download.
  • Apple silicon. The local models need an Apple silicon Mac.

Other Petal guides

To try Petal, get Petal for Mac. On Windows, Petal for Windows is a separate free preview with local models.

Questions

What is a free Aqua Voice alternative for Mac?

Petal for Mac is free and open source under the MIT license. It transcribes speech on your Mac with downloaded models, has no word cap and no account, and pastes the text into the front app.

Does Aqua Voice work offline?

Aqua Voice's pricing page does not list an offline mode, and third-party reviews describe it as cloud only. It needs HTTPS and WebSocket access on port 443 to transcribe.

How many free words does Aqua Voice give?

At the time of writing, the Aqua Voice pricing page lists 2,000 words on the Free plan. Pro, Max, and Business remove the word limit. Earlier 2026 pages quoted 1,000 words, so check the current page.

Does Petal send my text anywhere?

Only if you choose a cloud cleanup model. By default, speech and Petal W1 cleanup run on your Mac. With Cloud Model on, Petal sends the transcript only to the provider that you set up with your own key.

AlternativesMac

Alternative to Apple Dictation on Mac

Need an alternative to Apple Dictation on Mac? Petal adds model choice, on-device Petal W1 cleanup, and per-app routing, and can still use Apple Speech.

AlternativesMac

MacWhisper alternative for dictation

Need a MacWhisper alternative for dictation? Petal is free and open source, does live dictation in any app, and transcribes dropped audio files on device.