Loading
FluidVoice is the better pick if you are on macOS 15+ and want a completely free, open-source dictation tool with sub-100ms live streaming output and a multi-engine backend. Typilot wins when you need a stable Windows or Linux app, an AI command layer that acts on any text without cloud keys, on-device meeting transcription with speaker diarization, ghost-text autocomplete, or a one-time lifetime licence.
Yes. Both run speech recognition on-device and insert text in any app without sending audio to a server. FluidVoice is a free GPLv3 macOS dictation tool (macOS 15+) with sub-100ms Parakeet streaming and a Fluid-1 AI cleanup layer. Typilot adds 27 keyboard AI command prefixes via local Ollama, ghost-text autocomplete, on-device meeting transcription with speaker diarization, and stable native apps for Windows and Linux, starting at $7.99/month or $199 lifetime.
Yes. The core FluidVoice ASR engine layer is free and open source under GPLv3 with no word cap and no account required. The Fluid-1 AI cleanup layer is included at no cost but is a privately maintained runtime that is not part of the open-source release. Typilot offers a 3-day free trial then starts at $7.99/month or $199 lifetime.
Not as a stable release. FluidVoice is primarily a macOS 15+ app (Apple Silicon and Intel). An unsigned Windows pre-release build exists but is not the stable product. iOS is on a waitlist. Typilot ships stable native apps for macOS, Windows, and Linux.
Yes, when you select the Whisper engine. The default Parakeet engine covers 25 languages at sub-100ms latency; the Nemotron engine covers approximately 40 languages. Switching to the Whisper engine in FluidVoice gives 99+ language coverage comparable to Typilot, at the cost of slightly higher batch-mode latency.