Robot Learning from Human Demonstration for Playing Ocarina

Junfeng Wei,Ziang Liu,Fuchun Sun,Bin Fang
DOI: https://doi.org/10.1007/978-981-99-0301-6_14
2023-01-01
Abstract:Nowadays the field of combining robots with music has received great attention. However, the current work often focuses on how to control the mechanical performance of music with established rules. We hope that the robot can play a musical instrument without being so monotonous and can have playing skills other than regularity as a human being. We designed an anthropomorphic robotic ocarina playing system. The movements and corresponding action parameters when the dexterous hand is playing are designed so that the system can read the digital score and automatically play according to the score. Through our pitch and intensity analysis algorithm, the audio is parsed into the pitch sequence and intensity sequence required for ocarina performance, and the dexterous hand can play the ocarina through these two sequences so that the ocarina can be imitated by only listening to the audio. Furthermore, a granularity method is proposed to evaluate the similarity of performances.
What problem does this paper attempt to address?