Musical genre segmentation is an abecedarian task in music information retrieval, with operations classifying from music recommendation to association in large databases. This delves into the use of convolutional neural networks (CNNs) to classify music genres utilizing the GTZAN dataset, which comprises 1000 tracks across 10 distinct genres. Each audio train is converted into Mel-frequency Cepstral Portions (MFCCs), an extensively used point in audio analysis. A CNN model is employed to capture original and global patterns in spectrogram representations. Also, a relative analysis with traditional machine literacy models similar as SVM, KNN, Naive Bayes, random forest, and logistic retrogression is performed. Our model reaches a training delicacy of 99.25 and a test delicacy of 94.50, significantly outperforming other machine literacy models. The evaluation criteria, similar as perfection, recall, F1-scores, demonstrate the superiority of the CNN model in landing audio patterns. The results indicate that deep literacy is largely effective for genre identification in music and has implicit for colorful operations in digital music services. Unborn work will concentrate on enhancing the model by using larger datasets and exploring more advanced infrastructures.