Patent №
US 8,050,397
Granted
2011-11-01
Filed 2006
Owner
CISCO TECHNOLOGY, INC.
Lab
—
AI components
1
speech
Assignment
Recorded
Dataset
AIPD
2023_r1 edition
Application
11615666
In one embodiment, a method for discriminating between a machine generated multi-tone signal and a simulated voice multi-tone signal is provided. The method may comprise comparing output profiles generated from sampled outputs of a plurality of filters. The plurality of filters may have a single input sample stream applied to them and each filter may be preset at a measured multi-tone frequency associated with an identified prospective multi-tone signal. Based on the comparison of the output profiles, an inequality degree between the output profiles is generated and compared to an inequality threshold, thereby to determine whether the input sample stream comprises a machine generated multi-tone signal.
AI classification
Ownership
CISCO TECHNOLOGY, INC.
assignment · 186730532
Assignors
READ, BROOKS STEVENS
On an employer assignment, the assignors are typically the inventors.