Optional on-device wake word activation for hands-free dictation

[kxmXX]

FluidVoice currently relies on keyboard or mouse shortcuts to start dictation.

I'd like to propose an optional wake word feature that allows users to start dictation simply by saying a phrase such as "Hey Fluid", without touching the keyboard.

Proposed behavior

• Users can enable or disable wake word detection in Settings.
• When enabled, FluidVoice listens for a predefined wake phrase using lightweight, on-device processing.
• When the wake phrase is detected, FluidVoice starts its existing dictation flow, just like the primary keyboard shortcut.
• Wake word detection pauses while dictation is active and resumes afterward.
• Existing keyboard shortcuts remain fully functional.

Implementation approach

I propose adding a dedicated WakeWordService and connecting it to the existing dictation lifecycle, rather than simulating keyboard events or introducing another dictation pipeline.

The first version would focus on macOS, local processing, reliability, and low idle CPU usage.

Questions for maintainers

Would you be interested in accepting this feature?

If so, would you prefer a fixed wake phrase for the initial implementation, or support for configurable phrases?

I'd be happy to work on the implementation once the scope is approved.

[tvcnet]

like, "Ok Google..."