Abstract: The research deals with deep learning-based, voice-operated traffic sign recognition to improve road safety. A deep learning-based model is presented for the recognition of traffic signs with CNN, trained on GTSRB Dataset with an identification and categorization precision of 95%. The developed system detects signs and, through audio warnings by speakers, assists drivers to make quick decisions. The system aims to mitigate accidents caused by missed or misinterpreted signage by alerting drivers to nearby traffic signs and rules. This approach has potential applications in both driver assistance systems and autonomous vehicles.

Keywords: CNN, Keras, Tensor flow, GTSRB Dataset, Traffic Signs.

| DOI: 10.17148/IARJSET.2025.12120