SPEAKYou commented this keyword
FluidVoice
An open-source Mac app that types what you say into any app, with on-device speech models and optional AI cleanup.
- macOS 15 (Sequoia) or later
- Apple Silicon Mac (Intel works with Whisper models)
- Setup
- 5 min
- Level
- Beginner
- Stars
- 11.9k
- License
- GPL-3.0
What it does
FluidVoice is a voice-to-text app for macOS. Press a hotkey, speak, and your words are typed into whatever app you are using. Speech models run on your Mac, so your voice and text stay on your machine unless you choose a cloud AI provider.
Command Mode lets you control your Mac by voice, and Write Mode writes or rewrites text in any text box. Optional AI enhancement cleans up formatting, either fully on-device with Fluid Intelligence or with your own OpenAI or Groq key.
Install it
- Install with Homebrew using the command below, or download the latest release from GitHub.
- Grant microphone and accessibility access when asked. Both are needed to dictate and type into other apps.
- Pick a global hotkey in settings.
- Go through onboarding and choose a voice model for your language and speed needs.
Copy this
brew install --cask fluidvoice
Use it
- Press your hotkey anywhere and start talking. A live preview shows the words as you speak.
- Select text in any app and use Write Mode to rewrite it by voice.
- Optional: download Fluid Intelligence during onboarding for on-device cleanup, or add an OpenAI, Groq or custom provider key.
Good to know
Mac only for now; iOS and Windows have a waitlist. A voice model needs about 1 GB of disk, and Fluid Intelligence about 3.5 GB more.
The app sends one anonymous activity signal per day, and detailed anonymous analytics are on by default. Turn them off in Settings, Share Detailed Anonymous Analytics. Voice, audio and text are never collected.
The developers warn that fluidvoice.org is not their site. Download only from GitHub Releases or altic.dev/fluid.
Want the next one first?
One new guide a week, straight to your inbox.
