Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
SUPERVISED LEARNING METHOD AND SYSTEM FOR EXPLICIT SPATIAL FILTERING OF SPEECH
Document Type and Number:
WIPO Patent Application WO/2024/005403
Kind Code:
A1
Abstract:
A supervised learning method and system for explicit spatial filtering of speech are disclosed. A supervised learning method for spatial filtering of speech performed by a beamformer learning system, according to one embodiment, comprises the steps of: receiving a multi-channel speech signal incident on a microphone array in an environment where reverberation exists in a neural network-based beamformer model, and a beam condition expressing a direction of interest; and outputting a request signal corresponding to the beam condition from the multi-channel speech signal by using the neural network-based beamformer model, wherein the neural network-based beamformer model may be trained so as to extract speech signals of azimuth and elevation angles set in the beam condition by using training data.

Inventors:
CHANG JOON-HYUK (KR)
CHOI JEONGHWAN (KR)
Application Number:
PCT/KR2023/008049
Publication Date:
January 04, 2024
Filing Date:
June 12, 2023
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
IUCF HYU (KR)
International Classes:
G10L21/02; G06N3/08; G10L15/02; G10L25/30
Foreign References:
KR20200145219A2020-12-30
KR20190108711A2019-09-25
KR20220022286A2022-02-25
KR20210017252A2021-02-17
Other References:
MENG YU; XUAN JI; BO WU; DAN SU; DONG YU: "End-to-End Multi-Look Keyword Spotting", ARXIV.ORG, 21 May 2020 (2020-05-21), XP081675979
CHOI JEONGHWAN, CHANG JOON-HYUK: "Supervised Learning Approach for Explicit Spatial Filtering of Speech", IEEE SIGNAL PROCESSING LETTERS, IEEE, USA, vol. 29, 1 January 2022 (2022-01-01), USA, pages 1412 - 1416, XP093121519, ISSN: 1070-9908, DOI: 10.1109/LSP.2022.3181971
Attorney, Agent or Firm:
YANG, Sungbo (KR)
Download PDF: