Analyzing quality clarinet sound using deep learning. A preliminary study.

When a music student begins training, one of the main problems encountered is the proper understanding of specific terms that teachers introduce as a way of analyzing the type of sound produced by the student. The goal of a music teacher is that their students improve the quality of the sound they are emitting, but not in all cases students understand and know how to apply the concepts that the teacher wants the instrument to be able to emit the expected sound. Any tool that allows students to distinguish between sound quality would be helpful. The work presented here is a preliminary study of the quality of the sound emitted by a clarinet using deep learning techniques. It presents a first approximation of what will become a software tool that will allow music students to see that the sound quality they are emitting is correct and in real time. With this type of tools music students will be able to understand and associate the concepts explained by the teacher in a simpler way, and will even serve as a guide to improve their learning when the teacher is not present.

DOI: 10.1109/SSCI.2017.8285322