Master of Science (MS)
Deep learning, lyric extraction, mood classification, music classification, natural language processing, tag extraction
Grouping music into moods is useful as music is migrating from to online streaming services as it can help in recommendations. To establish the connection between music and mood we develop an end-to-end, open source approach for mood classification using lyrics. We develop a pipeline for tag extraction, lyric extraction, and establishing classification models for classifying music into moods. We investigate techniques to classify music into moods using lyrics and audio features. Using various natural language processing methods with machine learning and deep learning we perform a comparative study across different classification and mood models. The results infer that features from natural language processing are a valuable information source for mood classification. We use methods such as term-frequency/inverse-document frequency, continuous bag of words, distributed bag of words and pre-trained word embeddings to connect lyrical features to mood classes. Different arrangements of the mood labels for music are explored and compared. We establish that features from lyrics with natural language processing methods demonstrate high levels of accuracy using CNNs. Our final model achieves an accuracyof 71% compared to existing methods using SVMs that achieve and accuracy of 60%.
Akella, Revanth, "MUSIC MOOD CLASSIFICATION USING CONVOLUTIONAL NEURAL NETWORKS" (2019). Master's Projects. 736.