For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
Automatic Music Transcription of Piano Music Using Shared Gaussian Process Latent Variable Model
Takeshi IMAMURA Tomoko MATSUI
D - Abstracts of IEICE TRANSACTIONS on Information and Systems (Japanese Edition)
Publication Date: 2017/10/01
Online ISSN: 1881-0225
Type of Manuscript: PAPER
automatic music Transcription, Shared Gaussian Process Latent Variable Model, Rank Order Filter,
Full Text(in Japanese): PDF(1.3MB)
>>Buy this Article
We propose a novel method of automatic music transcription for piano music which converts acoustic data into some musical notations. In the method, the relationship between acoustic and performance data is modeled using Shared Gaussian Process Latent Variable Model (SGPLVM) through assuming both data are generated from a common latent variable. Although many approaches have been studied, any methods have not achieved a sufficient performance and further improvement is demanded. We attempt to improve it by utilizing SGPLVM. We show that our method outperforms existing methods in the frame level evaluation with the MAPS database (F-measure: 0.78).