← all repositories
altic-dev/FluidVoice

An open-source dictation app for Mac that never phones home

FluidVoice brings fully offline speech-to-text and AI-enhanced formatting to any macOS text field, offering an open-source alternative to cloud dictation services.

4.9k stars Swift Image · Video · Audio
FluidVoice
Collecting fresh signals — velocity needs a few days of history.
collecting data…
star history

What it does

FluidVoice is a native macOS dictation app that transcribes speech to text in real time and injects the result into any text field across any application. It bundles multiple on-device speech models—including Apple Speech, Whisper, and NVIDIA’s Parakeet variants—so transcription happens locally without network requests. An optional local enhancement layer called Fluid Intelligence handles smart formatting, capitalization, and post-processing, again without shipping audio to a server.

The interesting bit

The core app is GPLv3 open source, but the “Fluid Intelligence” enhancement runtime is deliberately kept private so the developers can sustainably fund a free dictation tool. It is a rare case of a commercial-adjacent AI runtime subsidizing an open-source frontend, with everything still running on your own hardware.

Key highlights

  • Supports eight distinct speech model families, from lightweight Apple Speech to multilingual Nemotron and low-latency Parakeet, with model sizes ranging from built-in to roughly 670 MB.
  • Two distinct interaction modes: Write Mode for dictating into any app’s text field, and Command Mode for launching apps and triggering system actions by voice.
  • Fully local-first by default; optional cloud enhancement via OpenAI or Groq requires explicit API key entry and opt-in.
  • Notch-aware live transcription overlay, global hotkey capture, per-app prompt configuration, and optional local audio history with ZIP export.
  • Requires macOS 15 (Sequoia) and mostly Apple Silicon; Intel Macs are limited to Whisper models.

Caveats

  • Fluid Intelligence, the on-device AI enhancement layer, is not open source; the developers say they are keeping it private for now and that this may change in the future.
  • Windows, iOS, and Linux versions are listed as coming soon, so the app is currently macOS-only and largely Apple Silicon-only.

Verdict

Worth installing if you want private, offline dictation that works across any Mac app. Not for you if you are on Intel and unwilling to use Whisper, or if you need a Windows or mobile release.

Frequently asked

What is altic-dev/FluidVoice?
FluidVoice brings fully offline speech-to-text and AI-enhanced formatting to any macOS text field, offering an open-source alternative to cloud dictation services.
Is FluidVoice open source?
Yes — altic-dev/FluidVoice is open source, released under the GPL-3.0 license.
What language is FluidVoice written in?
altic-dev/FluidVoice is primarily written in Swift.
How popular is FluidVoice?
altic-dev/FluidVoice has 4.9k stars on GitHub.
Where can I find FluidVoice?
altic-dev/FluidVoice is on GitHub at https://github.com/altic-dev/FluidVoice.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.