"A Machine Learning Approach for Music Genre ...

"A Machine Learning Approach for Music Genre Classification"

Author Name : D.V. V. Brahmachari, G. Kusuma Harinadh, Mary Swaroopa D

ABSTRACT

The categorization of musical genres is a significant issue in the field of music information retrieval. In this paper, we provide a technique for categorizing songs' genres based on word embeddings produced by the Word2Vec model. The suggested approach uses NLP techniques for text preparation and lyric representation. For feature extraction, we employ the Bag- of-Words model, and a Support Vector Machine (SVM) classifier is used to categories the music.28372 song lyrics from seven different genres make up the dataset used in this study. To create high-quality word embeddings, we train the Word2Vec model on a sizable corpus of text. We then represent the lyrics of each song in the dataset using these embeddings. By eliminating stop words, stemming, and turning all the lyrics to stem less to lowercase the letters Using several assessment criteria, including accuracy, precision, recall, and F1-score, we assess the effectiveness of our suggested methodology. Our test findings demonstrate that the suggested technique outperforms other cutting-edge methods, achieving a high classification accuracy of 85%. In conclusion, the suggested methodology for categorizing songs into various genres of music using word embeddings produced by the Word2Vec model and NLP approaches is a promising solution. This approach may be used to improve the user experience for music enthusiasts and can be adapted to other music recommendation systems.