Title:
EXTRACTION DEVICE, EXTRACTION METHOD, LEARNING DEVICE, LEARNING METHOD, AND PROGRAM
Document Type and Number:
WIPO Patent Application WO/2022/149196
Kind Code:
A1
Abstract:
This learning device comprises a conversion unit, a combination unit, an extraction unit, and an updating unit. The conversion unit converts a mixed sound, the sound sources of respective components of which are known, into embedding vectors of the respective sound sources using a neural network for embedding. The combination unit combines the embedding vectors using a neural network for combination to obtain a combination vector. The extraction unit extracts a target sound from the mixed sound and the combination vector using a neural network for extraction. The updating unit updates the parameter of the neural network for embedding such that a loss function calculated on the basis of information relating to the sound sources of the respective components of the mixed sound and the target sound extracted by the extraction unit is optimized.
Inventors:
DELCROIX MARC (JP)
OCHIAI TSUBASA (JP)
NAKATANI TOMOHIRO (JP)
KINOSHITA KEISUKE (JP)
OCHIAI TSUBASA (JP)
NAKATANI TOMOHIRO (JP)
KINOSHITA KEISUKE (JP)
Application Number:
PCT/JP2021/000134
Publication Date:
July 14, 2022
Filing Date:
January 05, 2021
Export Citation:
Assignee:
NIPPON TELEGRAPH & TELEPHONE (JP)
International Classes:
G10L17/00; G10L17/04; G10L17/18; G10L21/0308; G10L25/30
Domestic Patent References:
| WO2020240682A1 | 2020-12-03 |
Attorney, Agent or Firm:
SAKAI INTERNATIONAL PATENT OFFICE (JP)
Download PDF: