FluidVoice

FluidVoice is a free, open source voice-to-text app for Mac users who want to dictate into email, documents, chat, terminals, and code editors. It transcribes speech locally and can turn rambling dictation into edited text before inserting it into the active app. It requires macOS 15 or later and supports Apple Silicon and Intel Macs, with Whisper providing Intel compatibility.

The optional Fluid Intelligence runtime runs Fluid-1, a custom-trained local model that removes filler, corrects capitalization, and adjusts formatting. Per-app prompts let dictation take a casual tone in Slack or a more formal one in Mail, and the app selects the matching profile when focus changes. FluidVoice itself uses the GPL-3.0 license; Fluid Intelligence is a separate, privately maintained runtime.

Speech models include Nemotron Speech, Parakeet TDT, Cohere Transcribe, Apple Speech, and Whisper. Language support depends on the model. CoreML and Metal accelerate transcription on Apple Silicon, and a live preview shows text as you speak. Write Mode supports dictating new content or rewriting selected text in place. Command Mode can launch apps, run shortcuts, and trigger system actions by voice.

Dictation and local enhancement work offline. Voice recordings and text stay on your Mac unless you choose a cloud AI provider such as OpenAI, Groq, or a custom service for post-processing. The local enhancement runtime doesn't require API keys. Optional recording history stores past dictations locally and supports ZIP export; anonymous analytics exclude audio, transcripts, and prompts.