This project presents an NLP-driven mood classifier for song lyrics, leveraging deep learning techniques to enhance music recommendation systems. The classifier achieves an accuracy of 77.08% using a Convolutional Neural Network (CNN) with word2vec embeddings, outperforming traditional lyric-based models.
- NLP-driven Mood Classifier: Utilizes CNN with word2vec embeddings to classify the mood of song lyrics.
- High Accuracy: Achieves 77.08% accuracy, surpassing traditional models.
- Model Comparison: Compares deep learning models (CNN w2v0, CNN w2v1) with traditional machine learning models (Naive Bayes, SVM).
- Dataset: Utilizes the Million Song Dataset to train and evaluate models.
- Natural Language Processing (NLP): For processing and analyzing song lyrics.
- Deep Learning: Convolutional Neural Networks (CNN) with word2vec embeddings.
- Traditional Machine Learning: Naive Bayes, Support Vector Machine (SVM).
- Python Libraries: Keras, Sci-kit Learn.
- Enhances music recommendation systems by accurately classifying the mood of songs.
- Advances NLP applications in the music industry.
This project demonstrates the effectiveness of deep learning techniques in improving the accuracy of mood classification from song lyrics, providing a valuable tool for music recommendation systems.