Adaptive Music Streaming: Combining CNN Facial Analysis with Snowflake based Music Categorization
K. Deepa, S. Pradeesh, M. Mathesh, S. Saravanakumar, R. D. Jonas Jefferey · 2025
This paper introduces an adaptive music streaming system that combines facial emotion analysis with dynamic music categorization to deliver a personalized and engaging auditory experience. The system uses a webcam to capture real-time facial images and applies the Haar Cascade algorithm to detect faces accurately and efficiently. Once a face is detected, the system employs a Convolutional Neural Network (CNN) enhanced with a Convolutional Block Attention Module (CBAM). This enhancement ensures improved accuracy by focusing on the most critical features of the captured facial expressions, enabling greliable identification of emotional states such as happiness, sadness, and anger. The identified emotions are then mapped to a Snowflake-based database, which categorizes music based on emotional attributes. By matching the user's emotional state with corresponding music categories, the system retrieves and streams songs that align with their current mood. This ensures an engaging and personalized listening experience, adapting to real-time emotional changes seamlessly. This design enables seamless integration between real-time emotion recognition and music streaming, making it easily accessible to users. This innovation represents a significant step forward in emotion-responsive multimedia applications. It not only improves the interaction between users and music streaming platforms but also establishes a foundation for future developments in adaptive and emotion-aware technologies.