JSAI2018

Presentation information

Oral presentation

General Session » [General Session] 13. AI Application

[2D4] [General Session] 13. AI Application

Wed. Jun 6, 2018 5:20 PM - 7:00 PM Room D (4F Cattleya)

座長:森川 幸治(パナソニック株式会社)

6:20 PM - 6:40 PM

[2D4-04] Development of a Jam Session System with Hatsune Miku That Learns User's Favorite Scat by Reinforcement Learning

〇Joji Tsuruda1, Natsuki Oka1, Kazuaki Tanaka1 (1. Kyoto Institute of Technology)

Keywords:music, reinforcement learning, jam settion system, interaction

Musicians often use jam sessions as a way to communicate with music. Many studies on jam session systems, which enable jam sessions between people and machines, have been made. In jam sessions, singers often use the scat singing. In our proposed system, a user plays keyboard and the system will reply in scat singing of Hatsune Miku according to the user's performance. We aim to make the system learn the user's preferred scat words by reinforcement learning through jam sessions. We implemented the learning system and confirmed that the system could learn the user’s preference in scat words.