	US 12,431,146 B2
	Systems and methods for covariance smoothing
David S. McGrath, Rose Bay (AU); Stefanie Brown, Lewisham (AU); and Juan Felix Torres, Redfern (AU)
Assigned to Dolby Laboratories Licensing Corporation, San Francisco, CA (US)
Filed by Dolby Laboratories Licensing Corporation, San Francisco, CA (US)
Filed on Mar. 14, 2024, as Appl. No. 18/605,733.
Application 18/605,733 is a continuation of application No. 17/632,225, granted, now 11,972,767, previously published as PCT/US2020/044670, filed on Jul. 31, 2020.
Claims priority of provisional application 63/057,533, filed on Jul. 28, 2020.
Claims priority of provisional application 62/881,825, filed on Aug. 1, 2019.
Prior Publication US 2024/0265927 A1, Aug. 8, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G10L 19/008 (2013.01); G10L 19/26 (2013.01); G10L 21/0208 (2013.01); G10L 25/18 (2013.01)

CPC G10L 19/008 (2013.01) [G10L 19/26 (2013.01); G10L 21/0208 (2013.01); G10L 25/18 (2013.01); G10L 2021/02082 (2013.01)]

15 Claims

1. A method of providing an audio service, the method comprising:

for a time-domain sequence of signal frames, obtaining an effective count of bins in a frequency band and a desired count of bins for the frequency band;

computing a forgetting factor for the frequency band as a ratio of the effective count to the desired count;

for each frequency band of a plurality of frequency bands, generating a respective smoothed band-specific covariance matrix for a present signal frame using a previously generated value of the respective smoothed band-specific covariance matrix for a previous signal frame relative to the present signal frame and the forgetting factor; and

generating an audio signal for the audio service using frequency domain representations of the time-domain sequence of signal frames and further using corresponding sets of the respective smoothed band-specific covariance matrices;

wherein the obtaining, computing, and generating are performed by a system including one or more computer processors.