Local-first Windows voice dictation tool. Hold a key, speak, and get polished text injected anywhere on Windows.

Designed a highly modular, decoupled architecture by defining clear trait contracts for audio capture, STT engine, cleanup rules, and text injection.
Integrated whisper.cpp via whisper-rs for fast, on-device speech-to-text processing without relying on external cloud APIs.
Engineered the Win32 API integration strategy for global hotkey detection and seamless text injection, utilizing modern tooling to handle the low-level hooks.
Language
Framework
AI / STT
Runs whisper.cpp models directly on-device, ensuring complete privacy with no cloud audio processing.
Hold Ctrl+Space from anywhere in Windows to start dictating instantly.
Uses Win32 SendInput and clipboard pasting to inject transcribed text into the currently focused window.
Highly decoupled Rust architecture allowing easy swapping of STT models, cleanup rules, and input sources.
Reliable microphone capture using cpal (WASAPI) with automatic resampling.
Conservative regex and rules engine to polish transcribed text before injection.





I'm actively seeking opportunities to join a great team and contribute to exciting products. Drop me a message!