KebiControl

Voice control for Windows. Made by KebiLab.

How to use

1. Run KebiControl.exe.
2. Click the gear icon (top-right) to open Settings.
3. In Settings:
   • Распознавание речи (Whisper) — paste your OpenAI key (or any Whisper-compatible endpoint) for speech-to-text.
   • Нейросеть (LLM) — pick a provider (OpenCode Go, OpenAI, Anthropic, Google Gemini, Mistral, Groq, DeepSeek, xAI, or your own), choose a model, paste the API key. Keys are encrypted with Windows DPAPI.
   • Click Сохранить.
4. Back on the main window:
   • Click the big Говорить button and speak. The app will listen, recognize, execute the command and answer with voice.
   • Or type a command in the input and press Enter.
   • Or click a quick action: Pause, Screenshot, Quieter, Louder.
5. Wake word (optional). In Settings, tick "Пробуждение по голосу" and set the wake phrase (default: кеби). The mic will then stay on, and when the app hears the phrase it will start a full STT cycle automatically. Detection uses short probes against the same Whisper endpoint you configured for STT.
6. Switch language with the Русский / English button.
7. Switch theme with the Тёмная / Светлая button.

Voice pipeline:

microphone (cpal, 16 kHz)
  -> STT (Whisper HTTP, your key)
  -> text
  -> local parser (regex rules) or LLM fallback
  -> Command
  -> action executor (Win32 / PowerShell)
  -> voice reply (SAPI via PowerShell)

Build

git clone https://github.com/KebiLab/kebicontrol.git
cd kebicontrol
cargo build --release
.\target\x86_64-pc-windows-gnu\release\kebicontrol.exe

License

Apache 2.0. See LICENSE.

Made by KebiLab