Classical Music Composer Classifier

●What is the problem you are solving? We are trying to create a machine learning model that can accurately classify classical music by its composer.

● Why did you choose this problem? It's hard for apps like Shazam to accurately identify classical pieces, so we decided to challenge ourselves to see if we could create a machine learning model that could predict the composer's more accurately.

● Briefly explain the models you experimented with to solve the problem. We played around with CNN's, and we had experimented a lot on how to best convert a raw audio file into a image that had enough features for classification, such as midi, pianorolls, spectrograms, mel-spectrograms, etc. We also had to play around with different hidden layer setups in order to find a more accurate model.

● What is the result you are the most satisfied with (accuracy, other metrics, etc)? We are impressed with the training validation loss graph.

● If you were to continue this project, what would you explore doing / improve? We would want to expand the dataset, as the current dataset has only 10 labels, and is very heavily skewed towards one composer.

Built With

• cnn
• python