MULTI-CHANNEL-BASED INTEGRATED NOISE AND ECHO SIGNAL CANCELLATION DEVICE USING DEEP NEURAL NETWORK

WIPO Patent Application WO/2022/158912

A1

A multi-channel-based integrated noise and echo signal cancellation device using a deep neural network according to an embodiment may include: a plurality of mic-encoders for receiving a plurality of mic input signals which include echo signals, noise signals, and speech signals of utterers, and respectively converting the plurality of mic input signals into a plurality of pieces of conversion information and outputting the plurality of pieces of conversion information; a channel conversion unit for compressing the plurality of pieces of conversion information and thereby converting same into first input information having the size of a single channel, and outputting the first input information; a far-end signal encoder for receiving a far-end signal, converting the far-end signal into second input information, and outputting the second input information; an attention unit for applying an attention mechanism to the first input information and the second input information to output weight information; a trained first artificial neural network having, as input information, third input information that is aggregate information of the weight information and the second input information, and having, as output information, first output information including mask information for estimating the speech signal from the second input information; and a speech signal estimation unit for outputting an estimated speech signal obtained by estimating the speech signal on the basis of the first output information and the second input information.

CHANG JOON HYUK (KR)
PARK SONG KYU (KR)

PCT/KR2022/001164

July 28, 2022

January 21, 2022

Click for automatic bibliography generation  

IUCF HYU (KR)

G10L21/02; G06N3/08; G10L19/008; G10L21/0216; G10L25/30; H04R3/00; H04R3/02

KR20200115107A	2020-10-07
US20180040333A1	2018-02-08
KR102316712B1	2021-10-22

SEO HYEJI, LEE MOA, CHANG JOON-HYUK: "Integrated acoustic echo and background noise suppression based on stacked deep neural networks", APPLIED ACOUSTICS., ELSEVIER PUBLISHING., GB, vol. 133, 1 April 2018 (2018-04-01), GB , pages 194 - 201, XP055952259, ISSN: 0003-682X, DOI: 10.1016/j.apacoust.2017.12.031
GUILLAUME CARBAJAL; ROMAIN SERIZEL; EMMANUEL VINCENT; ERIC HUMBERT: "Joint NN-Supported Multichannel Reduction of Acoustic Echo, Reverberation and Noise", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 27 July 2020 (2020-07-27), 201 Olin Library Cornell University Ithaca, NY 14853 , XP081704527, DOI: 10.1109/TASLP.2020.3008974
HONGSHENG CHEN; TENG XIANG; KAI CHEN; JING LU: "Nonlinear Residual Echo Suppression Based on Multi-stream Conv-TasNet", ARXIV.ORG, CORNELL UNIVERSITY LIBRARY, 201 OLIN LIBRARY CORNELL UNIVERSITY ITHACA, NY 14853, 15 May 2020 (2020-05-15), 201 Olin Library Cornell University Ithaca, NY 14853 , XP081674382
PARK SONG-KYU, CHANG JOON-HYUK: "Multi-TALK: Multi-Microphone Cross-Tower Network for Jointly Suppressing Acoustic Echo and Background Noise", SENSORS, vol. 20, no. 22, 13 November 2020 (2020-11-13), pages 6493, XP055952907, DOI: 10.3390/s20226493

HAEUM PATENT & LAW FIRM (KR)

View/Download PDF      PDF Help

Previous Patent: METHOD AND APPARATUS FOR ENHANCING SECURITY OF MAC LAYER ENTITY IN NEXT-GENERATION MOBILE COMMUNICAT...

Next Patent: NOISE AND ECHO SIGNAL INTEGRATED CANCELLATION DEVICE USING DEEP NEURAL NETWORK HAVING PARALLEL STRUC...