Utilities
VCCM – Voice-Controlled Command Module
VCCM handles offline speech recognition and converts natural voice commands into actionable intent. Built using Vosk (STT) and LlamaCpp (local LLM), it supports contextual command interpretation.
What It Does
- Transcribes voice commands into text (offline)
- Uses NLP to extract intent and logic
- Sends control signals to Qognix based on speech
Example Use Cases
- Voice-triggered assistants on mobile robots
- Human-robot interaction in noisy environments
- Commands like “Watch for Mark” or “Alert me if someone falls”
Supported Components
| Type | Device Example | Notes |
|---|---|---|
| USB Microphone | Samson Go Mic / Blue Snowball | UAC class, cross-platform compatible |
| I2S Mic Module | ReSpeaker 2/4 Mic Array | Low-level mic arrays for embedded systems |
| HAT Audio Boards | Seeed ReSpeaker HAT for Pi / Jetson | Seeed ReSpeaker HAT for Pi / Jetson |
| Array Mic (USB) | MiniDSP UMA-16, Matrix Voice USB | Far-field voice capture with
beamforming |