How to use FluidVoice on a Mac: install it, choose a model, and what its modes do
By Ryleigh Newman · Published
FluidVoice is a free, open source dictation app for macOS, under GPLv3. To set it up, install it with Homebrew or from its GitHub releases, grant microphone and accessibility access, set a global hotkey, and pick a speech model during onboarding. It needs macOS 15 or later. Every model runs on Apple silicon, and Intel Macs are supported through the Whisper models.
I built YapToText, a dictation app for the Mac, so read what I say about FluidVoice with that in mind. Every fact about FluidVoice here comes from its README on GitHub, read on October 11, 2026.
This 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: 9 sections
Install it
brew install --cask fluidvoice- Or download the latest release from its GitHub Releases page.
- Grant microphone and accessibility access. Both are required for dictation and for typing into other apps.
- Set a global hotkey in settings.
- Go through onboarding and choose a speech model for your language.
Get it from GitHub or altic.dev/fluid. The README warns that fluidvoice.org is not affiliated with the project. It names fluidvoice.com and fluidvoice.app as its other official domains.
Choose a speech model
From FluidVoice's README, read October 11, 2026.
| Model | Best for, by its README | Languages | Download | Runs on |
|---|---|---|---|---|
| Nemotron Speech 3.5 | Streaming multilingual dictation | About 40 | About 670 MB | Apple silicon |
| Nemotron 3.5 Multilingual | Higher-accuracy multilingual dictation | About 40 | About 530 MB | Apple silicon |
| Parakeet Flash, in beta | Lowest-latency live English | English | About 250 MB | Apple silicon |
| Parakeet TDT v3 | Fast default multilingual dictation | 25 | About 500 MB | Apple silicon |
| Parakeet TDT v2 | English only | English | About 500 MB | Apple silicon |
| Cohere Transcribe | High-accuracy multilingual dictation | 14 | About 1.4 GB | Apple silicon |
| Apple Speech | No download | The system's languages | Built in | Apple silicon and Intel |
| Whisper, Tiny to Large | Broad compatibility, including Intel Macs | Up to 99 | About 75 MB to 2.9 GB | Apple silicon and Intel |
Parakeet TDT v3 is described as the fast default. Apple Speech is the choice if you want to start with no download at all, and Whisper is the one that works on an Intel Mac.
What the modes do
- Dictation. The hotkey captures your voice from any app, and the text is inserted through the accessibility APIs. A live preview overlay shows the words as you speak.
- Write Mode. Writes or rewrites text in any text field. Select text to rewrite it, or dictate new content in place.
- Command Mode. Controls the Mac by voice: launch apps, run shortcuts, trigger system actions.
- Per-app configuration. Assign different prompt sets to different apps. It is optional.
Cleanup: Fluid Intelligence or a provider
AI enhancement is optional and comes two ways. Fluid Intelligence is a local model for formatting and post-processing that runs on the Mac. The README says it is a separate runtime that is privately maintained, not part of the open source code, and it is about 3.5 GB to download. The other way is a cloud provider with your own API key, stored in the macOS Keychain. With neither, dictation still works from the speech model alone.
What it needs
- macOS. 15.0 Sequoia or later.
- Chip. Apple silicon for all models. Intel Macs through Whisper, from version 1.5.1.
- Disk. About 1 GB for a voice model, and about 3.5 GB more for Fluid Intelligence.
- Other platforms. The README says iOS and Windows are on the way, with a waitlist.
Privacy and analytics
The README calls FluidVoice local-first: voice, audio and transcribed text do not leave the Mac unless you opt in to a cloud AI provider. It does send analytics.
- Always. One anonymous activity signal per day, uploaded weekly: a random installation ID, the date, the app version and a macOS platform label.
- By default, and switchable. Detailed anonymous analytics: feature and model usage totals, onboarding progress, and model download outcomes. Turn it off under Settings, Share Detailed Anonymous Analytics.
- Not collected. Voice, audio, transcribed text, selected text, prompts, window titles, file paths, clipboard or typed content.
Turning detailed analytics off still leaves the weekly activity batch.
FluidVoice and YapToText
FluidVoice from its README, read October 11, 2026. YapToText from its own settings.
| FluidVoice | YapToText | |
|---|---|---|
| Price and license | Free, GPLv3 | Free, GPL-3.0 |
| Needs | macOS 15 or later. Intel through Whisper | macOS 14 or later, Apple silicon only |
| Install | Homebrew or GitHub releases | The Mac App Store |
| Speech models | Eight families to choose from | Whisper Large v3 Turbo, included |
| Local cleanup | Fluid Intelligence, an optional 3.5 GB download, privately maintained | A cleanup model inside the app, on by default |
| Controls the Mac by voice | Command Mode | No |
| Analytics | A weekly anonymous activity signal, plus optional detail | None |
FluidVoice is the better fit for an Intel Mac, for choosing among many speech models, or for Command Mode. YapToText fits if you want an App Store install with nothing to download and no analytics. The FluidVoice alternative page goes deeper.
This 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.
The short version
- Install. brew install --cask fluidvoice, or GitHub releases.
- Set up. Two permissions, a hotkey, a model.
- Model. Parakeet TDT v3 by default, Whisper on Intel, Apple Speech for no download.
- Cleanup. Fluid Intelligence on the Mac, or your own provider.
- Analytics. A weekly anonymous signal, with optional detail you can switch off.
If FluidVoice changes these steps, open an issue on GitHub and I will update this page. Every one gets read.
This 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.
Related
Sources
- https://github.com/altic-dev/FluidVoice (checked 2026-10-11)
- https://altic.dev/fluid (checked 2026-10-11)
- https://github.com/altic-dev/FluidVoice/releases (checked 2026-10-11)
- https://github.com/ryleighnewman/YapToText (checked 2026-10-11)
- https://apps.apple.com/us/app/yaptotext/id6786382289 (checked 2026-10-11)
Questions
How do I install FluidVoice?
Run brew install --cask fluidvoice, or download the latest release from its GitHub page. Then grant microphone and accessibility access and set a hotkey.
Is FluidVoice free?
Yes. Its README says it is free and open source under GPLv3.
Does FluidVoice work on Intel Macs?
Yes, through the Whisper models, from version 1.5.1. Its other models need Apple silicon.
What is Fluid Intelligence?
An optional local AI model for formatting and post-processing that runs on the Mac. The README says it is privately maintained and about 3.5 GB.
Does FluidVoice collect data?
It sends one anonymous activity signal per day in a weekly batch, and detailed anonymous analytics unless you switch them off. It says voice, audio and text are not collected.