Musical Beat Recognition using Machine Learning

In today's digital landscape, the majority of activities have transitioned to digital platforms. Music, as a widely cherished form of art, is both learned and taught by numerous individuals, leading to a surge in interest in music information retrieval (MIR) and its applications, fueled by advancements in machine learning technology. While applications like genre recognition, song identification, automatic score generation, and music transcription have seen significant development, there remains a gap in research concerning musical instrument identification. This paper presents a novel approach employing machine learning models in conjunction with Mel-frequency-cepstral-coefficients to extract audio features from raw datasets directly. The study utilizes a dataset comprising 600 audio files encompassing sound samples from six distinct instruments for training and instrument prediction purposes.