FluidVoice

What is FluidVoice?

FluidVoice is a free, open-source voice-to-text application for macOS that combines on-device AI with a polished user experience. Unlike traditional dictation tools that either stop at raw transcription or send your data to the cloud, FluidVoice pairs a native macOS app with Fluid-1, a local AI model that cleans up rough dictation, adapts tone per application, and formats text with proper casing, dates, names, and numbers—all without any data leaving your Mac.

Key features include hotkey activation (press the Option key), an overlay interface, local speech models, and the ability to type into any app—email, docs, chat, terminal, or code editor. Three modes—Write Mode, Command Mode, and Direct Dictation—let you switch context instantly. Write Mode polishes your speech into well-formatted text, Command Mode executes actions, and Direct Dictation outputs raw transcription. FluidVoice supports over 40 languages across models like Nemotron and Whisper, and leverages CoreML and Metal for low-latency, battery-efficient transcription.

Benefits include speaking approximately 3.7x faster than typing, zero data leaving your device (no Wi-Fi required), and adaptive tone customization—you can set prompts per app so dictation fits the context, from casual chat to professional emails. Use cases range from drafting documents and coding to sending quick messages or controlling your Mac hands-free.

Technical details: FluidVoice requires macOS 15.0+ and works on both Apple Silicon and Intel Macs. It is licensed under GPLv3, with community-visible code on GitHub. The app offers flexible AI provider options—local Fluid-1, OpenAI, Groq, or custom providers. With over 100,000 downloads and a growing community, FluidVoice is a powerful, privacy-first alternative to cloud-based dictation services.

Use Cases