Control your stream with your voice

Say party mode and the lights flash. Runs entirely on your PC - nothing is uploaded.

The desktop app can listen for spoken commands and act on them - flash the lights, switch an OBS scene, play a sound, pop an overlay. The speech model runs on your own computer: your microphone audio is transcribed locally and discarded, never uploaded, recorded or sampled. The only thing that touches the internet is a one-time download of the model itself.

1. Open the desktop app and go to the Voice tab.
2. Turn on voice control.
3. Download whisper.cpp and point GriddleBot at whisper-cli.exe - it is free and open source.
4. Download a model. Base at about 140 MB is the one to pick; Tiny is faster and rougher, Small is more accurate and heavier.
5. Choose how it listens: Press-to-talk uses a hotkey, Wake word listens for hey griddle all the time.
6. Pick your microphone, then adjust input sensitivity and the listen window.
7. Add a voice command: a phrase like party mode, and the action it should run.

Press-to-talk is the setting to start with. Wake word means the app is listening continuously, which costs CPU on the machine that is also encoding your stream, and it will occasionally trigger on something you said to chat.

On a Lights command you can name the colour as you speak - lights blue, lights rainbow - and it overrides the configured colour for that one firing. Whisper handles accents well and detects the language on its own.

All setup guides  ·  Install and pair the desktop app  ·  Put a macro on a Stream Deck button  ·  Control OBS from the dashboard

GriddleBot home  · Features  · Download  · Pricing  · Discover