Portal de Periódicos da CAPES

The research about editing system of performance information for player piano. -Develop inference methods using machine learning -

2023; Volume: 28; Linguagem: Inglês

10.5954/icarob.2023.os13-9

2435-9157

Takaaki Ueno, Sakmongkon Chumkamon, Eiji Hayashi,

Cultural and Historical Studies

In 1996, Professor Eiji Hayashi of this laboratory developed an automatic piano performance system that aims to reproduce performance skills equivalent to those of humans.¹However, this system does not have a function to infer performance expressions such as the strength and weakness of notes, and is not capable of expressing emotions such as intonation for each phrase, as is done by actual human performers.Therefore, the goal of this research was to reproduce the performances of world-famous pianists under the objective of "pursuit of more human-like performance expression.In the future, when this research matures and the human expression of the piano is clarified mathematically, the data will be used to reproduce human performances, which will fulfill the original purpose of this research.In this paper, we attempted to reproduce "Prelude Op.4 No.10" composed by Fryderyk Franciszek Chopin, using the worldfamous pianist "Vladimir Davidovich Ashkenazy" as the research subject.