	US 12,106,770 B2
	Sound model generation device, sound model generation method, and recording medium
Sakiko Mishima, Tokyo (JP); Yu Kiyokawa, Tokyo (JP); Takahiro Toizumi, Tokyo (JP); and Kazutoshi Sagi, Tokyo (JP)
Assigned to NEC CORPORATION, Tokyo (JP)
Appl. No. 17/623,316
Filed by NEC Corporation, Tokyo (JP)
PCT Filed Jul. 4, 2019, PCT No. PCT/JP2019/026659 § 371(c)(1), (2) Date Dec. 28, 2021, PCT Pub. No. WO2021/001998, PCT Pub. Date Jan. 7, 2021.
Prior Publication US 2022/0358953 A1, Nov. 10, 2022
Int. Cl. G06F 15/00 (2006.01); G10L 15/06 (2013.01); G10L 25/78 (2013.01)

CPC G10L 25/78 (2013.01) [G10L 15/063 (2013.01)]

9 Claims

1. A sound model generation device comprising:

at least one memory storing instructions; and

at least one processor configured to access the at least one memory and execute the instructions to:

generate a concatenated feature by linking a plurality of features of a sound signal in a channel direction, wherein the concatenated feature is a multichannel feature including the plurality of features of the sound signal as a plurality of channels, and the plurality of features of the sound signal are training data;

learns the concatenated feature; and

generate a sound model for distinguishing a sound event from the sound signal.