The motivation for this project was to apply principles from various multiclass classification techniques to a problem with a wide variety of potential output classes. We also wanted a problem with a large potential training dataset. Drawing on our interest in music, the objective we decided on was to find and implement a classification technique that could identify, from an audio file, individual musical notes. This posed a fascinating challenge due to the non-linear nature of sound wave features, as well as determining how to effectively represent this information for machine learning models.
