MusicMood

A machine learning approach to classify music by mood based on song lyrics.

This project is about building a music recommendation system for users who want to listen to happy songs. Such a system can not only be used to brighten up one's mood on a rainy weekend; especially in hospitals, other medical clinics, or public locations such as restaurants, the MusicMood classifier could be used to spread positive mood among people.

Links

• The web application
• The data collection IPython notebook
• The initial model training IPython notebook
• The updated model training with white lists IPython notebook
• Experiments with Random Forests IPython notebook
• An article about my experiences with this project
• A keynote presentation about this project
• A more technical report on arXiv

Sections

• Dataset Summary
• Exploratory Data Analysis
• Results

Dataset Summary

back to top

• A 10,000-song subset was downloaded from the Million Song Dataset.
• Lyrics were automatically downloaded from LyricWikia and all songs for which lyrics have not been available were removed from the dataset.
• An English language filter was applied to detect and remove all non-English songs.
• The remaining songs were randomly subsampled into a 1000-song training dataset and 200-song validation dataset.

Exploratory Data Analysis

back to top

Results

back to top