FluidVoiceFrequently asked questions
What people ask before they install. Anything else? Ask in GitHub Discussions.
Is FluidVoice free?
Yes. FluidVoice is free and open source, with no paid tiers.
Which devices does it run on?
Mac is available now. Windows is in beta, the iOS beta comes through TestFlight, Linux is in early beta and Android is in progress. Features and models vary by device.
What does my Mac need?
macOS 15 Sequoia or later, on Apple Silicon or Intel. Intel Macs use the Whisper models. FluidVoice asks for microphone access to record, and accessibility permission to type into other apps.
Does my dictation leave my device?
Not by default. Your audio and text are processed on your device. If you turn on an optional cloud provider, the text that feature needs is sent to that provider. Optional usage analytics are separate and never include what you say.
Does it work offline?
Yes. Dictation runs on local speech models, so it keeps working with no connection. Only the optional cloud providers you choose to turn on need the internet.
I have a base M1 MacBook Air. Where should I start?
Start with Fluid Speech Mini for transcription and Fluid Ripple for cleanup. Together they are under 1 GB to download and use under 1 GB of memory, though memory use varies with recording length.
Which Fluid model should I choose?
Fluid Wave is the largest model. Fluid Splash is under 1.5 GB, and Fluid Ripple is a 600 MB download for lighter machines. All three clean up your text on device through Fluid Intelligence on Mac; pair any of them with Fluid Speech Mini.
Which speech models can I use?
On Mac: Fluid Speech Mini, Nemotron Speech 3.5 (Ultra Fast Low Latency and Multilingual), Parakeet Flash (beta), Parakeet TDT v3 and v2, Cohere Transcribe, Apple Speech, and Whisper from Tiny to Large.
How many languages does it support?
It depends on the model. Nemotron Speech 3.5 covers about 40 languages, Parakeet TDT v3 covers 25, Cohere Transcribe 14 and Whisper up to 99. Parakeet Flash and Parakeet TDT v2 are English only, and Apple Speech follows your system languages.
What do the analytics collect?
Analytics are optional and anonymous. FluidVoice never collects your voice, raw audio, transcribed text or prompts.
What license does FluidVoice use?
GPLv3, from 2026-02-23 onward. Versions before that date used the Apache License 2.0.