US 11,894,012 B2
Neural-network-based approach for speech denoising
Changxi Zheng, New York, NY (US); Ruilin Xu, New York, NY (US); Rundi Wu, New York, NY (US); Carl Vondrick, New York, NY (US); and Yuko Ishiwaka, Tokyo (JP)
Assigned to The Trustees of Columbia University in the City of New York; and SoftBank Corp., Tokyo (JP)
Filed by The Trustees of Columbia University in the City of New York, New York, NY (US); and SoftBank Corp., Tokyo (JP)
Filed on May 19, 2023, as Appl. No. 18/320,206.
Application 18/320,206 is a continuation of application No. PCT/JP2021/027243, filed on Jul. 20, 2021.
Claims priority of provisional application 63/116,400, filed on Nov. 20, 2020.
Prior Publication US 2023/0306981 A1, Sep. 28, 2023
Int. Cl. G10L 21/0232 (2013.01); G10L 25/30 (2013.01); G10L 21/0216 (2013.01); G10L 25/18 (2013.01)
CPC G10L 21/0232 (2013.01) [G10L 25/30 (2013.01); G10L 25/18 (2013.01); G10L 2021/02168 (2013.01)] 8 Claims
OG exemplary drawing
 
1. A method comprising:
receiving an audio signal representation;
detecting in the received audio signal representation, using a first learning model, one or more silent intervals with reduced foreground sound levels;
determining based on the detected one or more silent intervals an estimated full noise profile corresponding to the audio signal representation; and
generating with a second learning model, based on the received audio signal representation and on the determined estimated full noise profile, a resultant audio signal representation with a reduced noise level.