| Project/Area Number |
15KK0008
|
|---|
| Research Category |
Fund for the Promotion of Joint International Research (Fostering Joint International Research)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Entertainment and game informatics 1
|
|---|
| Research Institution | Nagoya Institute of Technology |
|---|
Principal Investigator |
SAKO SHINJI 名古屋工業大学, 工学(系)研究科(研究院), 准教授 (30396791)
|
|---|
| Research Collaborator |
Rigoll Gerhart ミュンヘン工科大学, ヒューマンマシンコミュニケーション研究所, 教授
Kwolek Bogdan AGH科学技術大学, 准教授
Merendez Rafael Ramirez ポンペイファブラ大学, 准教授
|
|---|
| Project Period (FY) |
2016 – 2017
|
| Project Status |
Completed (Fiscal Year 2017)
|
| Budget Amount *help |
¥12,090,000 (Direct Cost: ¥9,300,000、Indirect Cost: ¥2,790,000)
|
|---|
| Keywords | マルチモーダル実演奏データ / 楽譜追跡 / セグメンタル条件付き確率場 / 畳み込みニューラルネットワーク / 運指推定 / 演奏表情生成 / 画像、文章、音声等認識 / 感性情報学 |
|---|
| Outline of Final Research Achievements |
I worked on improving music score following method using acoustic signals.
We developed a new score following method that utilizes musical score information such as part information of percussion instrument and melodies in addition to basic note sequence that are used in the conventional method. I confirmed that it is possible to improve the accuracy of score following without impairing the real time performance by conducting experiment using the RWC music data set. I also worked on an image processing method to acquire finger shape movements of music performance. I developed the multi-modal data set of music performance. In the hand shape recognition method using the Convolution Neural Network, it has been confirmed that the expanding training data set using precise 3-dimensional hand model could improve the recognition accuracy for real images.
|
