Our capstone project involved developing a music genre prediction application that identifies the top three music genres for a user-defined audio clip, along with their corresponding confidence levels. We utilized the GTZAN audio dataset to train a convolutional neural network (CNN). Audio processing was conducted using Librosa, and model training was performed with TensorFlow. The user interfaces were developed using Tkinter for the desktop version and Flask for the web version. This project was collaboratively completed by Juan Pablo Duque Ochoa, Jonathan Louangrath, and Efe Torunoglu.