This repository is inspired from Million Song Dataset Challenge from Kaggle. The Million Song Dataset Challenge aims at being the best possible offline evaluation of a music recommendation system. Reference : https://www.kaggle.com/c/msdchallenge/data
How do we label songs?90’s song or 80’s song. Searching for a song within genre when be- comes easier when we have an idea what era a song belongs to. Year prediction has not been studied very much. There is huge untapped potential if we could use year prediction in recommendation system. So in short, an interesting problem to work on.
Features being used are timbre average and covariance of every song, target is the year. For more description about the dataset refer to the url http://millionsongdataset.com/pages/getting-dataset/
Thierry Bertin-Mahieux, Daniel P.W. Ellis, Brian Whitman, and Paul Lamere. The Million Song Dataset. In Proceedings of the 12th International Society for Music Information Retrieval Conference (ISMIR 2011), 2011.