FluidVoice1.6.9

Hold a keyboard shortcut and speak into FluidVoice to place polished text directly into the active field, turning rough, rambling dictation into writing ready for email, documents, chats, terminals, and code editors. A local speech model first produces a raw transcript, while the custom-trained Fluid-1 model removes filler, corrects capitalization, structures the text, and adapts its tone to the app you are using. Speaking is reported to be about 3.7 times faster than typing, so ideas can appear on screen as quickly as you say them.

FluidVoice keeps the entire pipeline on your Mac: your voice is not sent to a server, and transcription continues without an internet connection. Write Mode, Command Mode, and Direct Dictation let you switch between composing text, issuing commands, and capturing speech, while support for more than 40 languages spans models including Nemotron and Whisper. CoreML and Metal provide low-latency processing with attention to battery efficiency, particularly on Apple Silicon Macs.

Developers, multilingual teams, writers, and anyone who works across multiple text fields can use the same shortcut throughout their workflow. The application is open source and distributed as freeware in the System Tools category for macOS, with version 1.6.9 supporting macOS 15 Sequoia and later on both Intel and Apple Silicon Macs. FluidVoice has surpassed 100,000 downloads and is designed around local, on-device AI rather than cloud dictation.