Automated Music Genre Classification Using Modified MobileNet Deep Learning Model

This research work intends to design and develop the music genre classification process for any given audio file and extract features from audios to understand the audio graph. The entire learning process has been conducted using a pre-trained Convolutional Neural Network (CNN) and the MobileNetV2 model with fine-tuning. Music Genre Classification, which categorizes music based on composition, such as the beat, is a focal point of this work. In recent years, music genre classification has become more approachable and popular with more genres emerging in the music sector that can now be automatically classified. The automated classification technology is also a suitable solution to reduce the effort involved in selecting music files manually. This work reviews a recent topic in the field of identification called music genre classification. The proposed model aids in identifying music files and creating a comprehensive playlist of songs. Manual selection of music files is not only difficult but also time-consuming. The modified MobileNetV2 model yields outstanding performance in terms of accuracy, precision, recall, and F1-score.