Title:
MACHINE LEARNING METHOD AND MACHINE LEARNING DEVICE
Document Type and Number:
WIPO Patent Application WO/2019/235633
Kind Code:
A1
Abstract:
A machine learning device comprising: a component enhancement unit 21 for inputting an acoustic signal X including a first component and a second component in a neural network N and thereby generating an acoustic signal Y in which the first component has been enhanced; a signal processing unit 22 for processing the acoustic signal Y and thereby generating an acoustic signal Z; and a learning processing unit 30 for training the neural network N in accordance with an evaluation index calculated from the acoustic signal Z.
Inventors:
NAKAJIMA HIROAKI (JP)
TAKAHASHI YU (JP)
TAKAHASHI YU (JP)
Application Number:
PCT/JP2019/022825
Publication Date:
December 12, 2019
Filing Date:
June 07, 2019
Export Citation:
Assignee:
YAMAHA CORP (JP)
International Classes:
G06N3/08
Foreign References:
| JP2018031967A | 2018-03-01 |
Other References:
KOIZUMI, YUMA ET AL.: "DNN-Based Source Enhancement to Increase Objective Sound Quality Assessment Score", IEEE/ACM TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING, vol. 26, no. 10, 30 May 2018 (2018-05-30), pages 1780 - 1792, XP055661738, Retrieved from the Internet
Attorney, Agent or Firm:
SHIN-EI PATENT FIRM, P.C. (JP)
Download PDF:
Previous Patent: BELT ACCESSORY, AND BELT EQUIPPED WITH BELT ACCESSORY
Next Patent: MOLDING EQUIPMENT
Next Patent: MOLDING EQUIPMENT