In order to classify our songs, we implemented algorithms to filter the noise out of the song (wavelet denoising), detect the beat of the signal, classify the key signature of the song using two separate methods, and classify the songs based on a bag-of-tones model (MFCC). These algorithms are described in more detail in each of their respective pages that can be accessed by the banner at the top of the page or if you press on the buttons below. The results of these algorithms are discussed in the Results section.
Wavelet Denoising Algorithm: Removes unwanted noise from audio input
Beat Detection Algorithm: Outputs a tempo of an input file in beats per minute
Key Signature Classification: Outputs the key signature of an input file
Classifies a test input file into one of three genres using MFCC and KNN