ASTA is an intelligent edge-cloud system that converts natural voice commands into executable actions using speech recognition and Large Language Models. Designed for resource-constrained devices, the framework dynamically balances computation between local edge hardware and cloud-based AI services to achieve low latency, high accuracy, and efficient resource utilization.
The project focuses on practical AI deployment by combining automatic speech recognition, language understanding, and adaptive inference strategies within a scalable software architecture. ASTA demonstrates how intelligent systems can provide responsive, reliable voice interaction while operating efficiently across heterogeneous computing environments.