A Comparative Study on the Drum Sound Recognition Algorithms Based Deep Learning
Sang Wook Lee, Jae Hyuk Heo, Sung Taek Lee · 2021
Drums are difficult to recognize not only because the sounds overlap when many instruments are played simultaneously, but also the notes and beats vary depending on how they are played. This study compared the Recurrent Neural Network (RNN) algorithm that analyzes the sequence of time and the Conversion Neural Network (CNN) algorithm that analyzes data in image form to compare which machine learning algorithms show better performance for drum sound recognition. Through this, this research wanted to study which algorithm to recognize musical instruments is appropriate, and contribute to the area of Music Information Retrieval (MIR).