Title:
SOUND GENERATING METHOD USING MACHINE LEARNING MODEL, TRAINING METHOD FOR MACHINE LEARNING MODEL, SOUND GENERATING DEVICE, TRAINING DEVICE, SOUND GENERATING PROGRAM, AND TRAINING PROGRAM
Document Type and Number:
Japanese Patent JP2022122689
Kind Code:
A
Abstract:
To provide a sound generating method, a training method, a sound generating device, a training device, a sound generating program, and a training program with which natural voices can be easily acquired.SOLUTION: For sections of a musical note sequence consisting of a plurality of sections, input of a representative value of musical feature quantities is received by a receiving unit 12. A first feature quantity column according to the input representative values of the sections is processed by using a trained model M, and a sound data column corresponding to a second feature quantity column in which the feature quantities continuously change is generated by a generating unit 13.SELECTED DRAWING: Figure 2
More Like This:
| JPH05188932 | MUSICAL SOUND RECOGNITION DEVICE |
| WO/2021/021305 | OBTAINING A SINGING VOICE DETECTION MODEL |
| JPS5647096 | SOUND SCALE GENERATOR CIRCUIT |
Inventors:
SAINO KEIJIRO
DAIDO RYUNOSUKE
JORDI BONADA
MELRAIN BRAU
DAIDO RYUNOSUKE
JORDI BONADA
MELRAIN BRAU
Application Number:
JP2021020085A
Publication Date:
August 23, 2022
Filing Date:
February 10, 2021
Export Citation:
Assignee:
YAMAHA CORP
International Classes:
G10H1/00; G10L13/033; G10H1/06; G10L13/10
Attorney, Agent or Firm:
Masahiro Nakagawa
Go Sakane
Hideyuki Sawamura
Go Sakane
Hideyuki Sawamura