Smart AI Pocket for Personalized Assistant

Abstract

Today, people are too dependent on digital assistant services, which restricts their capacity to meaningfully and productively interact with computers. This research introduces Smart AI Pocket, a handheld AI assistant based on natural language processing (NLP), embedded technologies and artificial intelligence (AI). The system uses the ESP32 microcontroller and is connected to the LLM API and cloud services to process voice commands and deliver context-aware responses in real-time. Input via voice, processing of data, connection, conversion of speech to text, and output make up the system architecture of this system. By achieving a success rate of 100% for general knowledge, mathematical calculation, graded questions, and voice clearance, 80% for command recognition, and 60% for voice input in noisy conditions, functional testing demonstrates that all functions execute admirably. The results of the performance study reveal an average end-to-end latency of about 16.8 seconds, an accuracy rate of 77% for speech recognition under normal conditions and 30% in noisy environments, and a stability rate of 95% for the Wi-Fi connection. The results demonstrate that the pocket-sized personal assistant Smart AI Pocket was well-designed and executed.