AI Software IP for Microcontrollers. Dialogs, security, multimodal triggers — once cloud-only, now on microcontrollers.
Development of Voice & Vision AI modules for microcontrollers — compact systems that enable natural interaction entirely offline.
Projects focus on local speech recognition, on-device voice synthesis, sound-based event detection, and multimodal edge intelligence.
Core directions:
• Offline ASR – fast and private voice control for embedded systems
• TinyTTS – natural speech generation directly on hardware
• SoundGuard – detection of dangerous or critical sounds
• SmallLM – miniature language models for intent understanding
Recent work extends into vision-enabled modules for smart access control, wildlife monitoring, and hybrid audio-video sensing at the edge.
Goal: to make microcontrollers capable of listening, understanding, and responding — naturally and securely, without relying on the cloud.
