Mining Creative Methods with Automatic Music Generation
|時間：||2018-01-18 (Thu) 10:00 - 12:00|
In this talk, I would like to discuss how automatic music generation enriches our creation activities. With the recent revival of curiosity in artificial intelligence, automated generation of musical contents utilizing deep neural networks is in the spotlight. However, it is not still clear how they can benefit our creation activities, and one might get even radical that human creators need not to create new works any more. I will present a perspective from a technical point of view how automatic music generation can actually contribute to creation by human. I will show creation methods or tips for creation which human creators are eager to learn could be extracted and verified through the development of automatic content generation. As a proof of concept, I would like to present my recent projects in mining creation methods and automatically generating dance motion (MachineDancing), music (ChordSequenceFactory, AutoGuitarTab) and cover songs (Song2Quartet, Song2Guitar).
Satoru Fukayama (深山 覚) received his Bachelor degree in Earth and Planetary Physics in 2008, and both Master and Ph.D. in Information Science and Technology in 2013, from the University of Tokyo. He is currently a Senior Researcher at the National Institute of Advanced Industrial Science and Technology (AIST), Japan. His main interests are in the theory and applications of automated music generation which is leveraged by machine learning. He is also a composer and have studied composition with E. Kawasaki and K. Kunikoshi. His most popular work is the automatic composition system named "Orpheus" which generates songs from arbitrary Japanese lyrics input. It has been implemented as a web-based interactive system with around 50,000 songs automatically generated per year by the user, and it has been introduced in more than 13 television shows in Japan. He has received awards including IPSJ Yamashita SIG Research Award, Specially Selected Paper Award and several Best Presentation Awards from the Information Processing Society of Japan.