Intelligent Driver Monitoring System for Detecting Fatigue and Distraction through Speech Analysis

Driver drowsiness and distraction contribute to the majority of road accidents. This paper introduces speech-based intelligent driver monitoring systems using Wav2Vec2.0, fine-tuned on the “Deeply Nonverbal Vocalization Dataset”. The system achieved 88.9% classification accuracy in fatigue detection by means of non-verbal vocalizations, such as yawning. The non-intrusive, privacy-preserving nature of this design overcomes the limitations of current methods while offering the possibility of providing real-time alerts for improved road safety and fewer accidents.