FluidVoice is a high-performance, open-source voice dictation and transcription utility designed natively for macOS 15.0+. Engineered as a completely free alternative to proprietary tools like MacWhisper and SuperWhisper, FluidVoice combines an on-device speech-to-text pipeline with the custom Fluid-1 AI model. It allows users to dictate into any text box across operating system apps instantly, polishing grammar, punctuation, and structural formatting completely offline without exposing raw audio files or transcripts to the cloud.

• Fluid-1 Local Post-Processing — local intelligence layer that cleans up filler words, structures unstructured thoughts, adjusts text casing, and formats complex technical terminology on-device
• Context-Aware Adaptive Tone — automatically detects the active window or application (such as Slack, Mail, Xcode, or Cursor) and dynamically alters the transcription style from casual to formal or structured Markdown
• Extensive Local Model Lineup — integrates optimized routing across a variety of speech models including Nemotron Speech 3.5, Parakeet Flash, Parakeet TDT, Cohere Transcribe, Apple Speech, and Whisper variations
• System-Wide Native Integration — features hardware-accelerated processing via CoreML and Metal execution layers, offering sub-100ms perceived latency and global hotkey triggers compatible with any code editor, browser, or terminal terminal pipeline
• Predefined Workflows & Modes — shift context seamlessly by utilizing dedicated system profiles like Write Mode, Command Mode, and Direct Dictation alongside support for more than 40 languages

While MacWhisper and SuperWhisper are popular choices in our utility directory, FluidVoice sets itself apart by being 100% free and open-source (GPLv3). Additionally, instead of stopping at basic transcription, FluidVoice runs the local Fluid-1 model to rewrite, format, and adjust the tone of your text automatically to match whatever app you are currently typing into.

No. FluidVoice operates entirely local-first. Both the speech recognition models and the Fluid-1 post-processing engine run natively on your Mac hardware, allowing you to use full-featured dictation with zero data leaving your machine and no mandatory accounts or telemetry.

Yes. While the default architecture relies entirely on offline hardware execution, FluidVoice includes an architectural framework that lets users optionally link custom cloud API providers, such as OpenAI, Groq, or customized endpoints if preferred.