Title:
MASK ESTIMATION DEVICE, MODEL LEARNING DEVICE, SOUND SOURCE SEPARATION DEVICE, MASK ESTIMATION METHOD, MODEL LEARNING METHOD, SOUND SOURCE SEPARATION METHOD, AND PROGRAM
Document Type and Number:
WIPO Patent Application WO/2019/163736
Kind Code:
A1
Abstract:
This mask estimation device for estimating mask information for specifying a mask used to extract the signal of a specific sound source from an input acoustic signal comprises: a conversion unit that converts the input acoustic signal to an embedded vector of a predetermined dimension using a learned neural network model; and a mask calculation unit that calculates the mask information by fitting the embedded vector to a mixed Gaussian model.
Inventors:
HIGUCHI TAKUYA (JP)
NAKATANI TOMOHIRO (JP)
KINOSHITA KEISUKE (JP)
NAKATANI TOMOHIRO (JP)
KINOSHITA KEISUKE (JP)
Application Number:
PCT/JP2019/005976
Publication Date:
August 29, 2019
Filing Date:
February 19, 2019
Export Citation:
Assignee:
NIPPON TELEGRAPH & TELEPHONE (JP)
International Classes:
G10L21/028; G10L21/0272
Other References:
CHEN, ZHUO: "DEEP ATTRACTOR NETWORK FOR SINGLE-MICROPHONE SPEAKER SEPARATION", PROCEEDINGS OF 2017 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, March 2017 (2017-03-01), XP081362980
HERSHEY, JOHN R.: "DEEP CLUSTERING: DISCRIMINATIVE EMBEDDINGS FOR SEGMENTATION AND SEPARATION", PROCEEDINGS OF 2016 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, March 2016 (2016-03-01), XP032900557
HERSHEY, JOHN R.: "DEEP CLUSTERING: DISCRIMINATIVE EMBEDDINGS FOR SEGMENTATION AND SEPARATION", PROCEEDINGS OF 2016 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, March 2016 (2016-03-01), XP032900557
Attorney, Agent or Firm:
ITOH, Tadashige et al. (JP)
Download PDF:
Previous Patent: COLORED RESIN COMPOSITION
Next Patent: ALLYLAMINE COPOLYMER AND METHOD FOR PRODUCING SAME
Next Patent: ALLYLAMINE COPOLYMER AND METHOD FOR PRODUCING SAME