US 12,321,432 B2
Computer implemented method
Clive David Summerfield, Ledbury (GB)
Assigned to FARX TECHNOLOGIES LIMITED, Malvern (GB)
Appl. No. 18/026,075
Filed by FARX TECHNOLOGIES LIMITED, Malvern (GB)
PCT Filed Sep. 14, 2021, PCT No. PCT/GB2021/052378
§ 371(c)(1), (2) Date Mar. 13, 2023,
PCT Pub. No. WO2022/053834, PCT Pub. Date Mar. 17, 2022.
Claims priority of application No. 2014436 (GB), filed on Sep. 14, 2020.
Prior Publication US 2023/0359719 A1, Nov. 9, 2023
Int. Cl. G06F 21/32 (2013.01); G06V 40/16 (2022.01); G06V 40/70 (2022.01); G10L 17/02 (2013.01); G10L 17/10 (2013.01)
CPC G06F 21/32 (2013.01) [G06V 40/171 (2022.01); G06V 40/172 (2022.01); G06V 40/70 (2022.01); G10L 17/02 (2013.01); G10L 17/10 (2013.01)] 15 Claims
OG exemplary drawing
 
1. A computer-implemented method of authenticating an identity of a user, the method comprising the steps:
acquiring a first data set representative of a voice of a user over a time interval between a first and second time, and providing the first data set as input to a computing device;
processing the first data set to extract acoustic parameters pertinent for both voice biometric and speech recognition analysis;
acquiring a second data set representative of a visual appearance of at least a portion of the user over the time interval between the first and second time, and providing the second data set as input to the computing device;
processing the second data set to extract user visual feature parameters pertinent for face recognition analysis;
fusing together of the first and second data sets at the parameter level to maintain a temporal synchronous over the time interval using a clock to ensure that the results of the voice biometric, speech recognition and face feature recognition analysis are generated synchronously;
comparing the first and second data sets with predetermined data sets relating to the voice and visual appearance of at least a portion of the specific user;
generating a confidence level in dependence of a relative correspondence of the first and second data sets with the predetermined data sets; and
authenticating the user as the specific user where the confidence level is above a predetermined value.