Music Genre Classification and Recognition using Improved Deep Convolutional Neural Network-DenseNet-II

Music genres are a group of keywords and descriptions that give high-level information about a specific part of music. However, classifying music into different genres is challenging due to insufficient information about the genre from the extracted features. To solve his problem, the Improved Deep Convolutional Neural Network-DenseNet-II (IDCNN-DenseNet-II) approach is proposed for enhancing the music genre classification and recognition of the music signals. The GTZAN dataset utilized for the classification is initially fed to Short Time Fourier Transform (STFT) technique. Then, the signals are preprocessed by the standardization method to scale the signals in the uniform range. At last, classification was performed by the proposed ID-CNN model to classify the various genres learned from the extracted relevant features. The evaluation results of the proposed method using performance metrics are Accuracy, Recall, Precision, and F1-score. The proposed approach attained a high accuracy with 98.2% of accuracy, 98.1% of precision, 97.9% of F1-score, and 98.0% of recall which is greater than the existing methods like Bidirectional-Long Short-Term Memory (Bi-LSTM), Deep learning Bag, 1D-resgated CNN-Transfer learning, CNN-Transformer, and Dual Parallel Attention-CNN.