# US 11,451,658 B1 · Systems and methods for proactive fraudster exposure in a customer service channel > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,451,658 B1 - Document ID: US11451658B1 - Status: Granted - Kind: B1 - Application number: 17/320,295 - Filed: 2021-05-14 - Granted: 2022-09-20 - Earliest claimed: 2021-05-14 - Assignees: Nice Ltd. - Applicants: Nice Ltd. - Inventors: Roman Frenkel; Gal Wolbe; Ofir Lavi - Web page: https://patentgrep.com/patent/US11451658B1 ## CPC - H04M 3/2281: {Call monitoring, e.g. for law enforcement purposes; Call tracing; Detection or prevention of malicious calls} - G10L 17/12: Score normalisation - H04M 3/42221: {Conversation recording systems (at the subscriber's set H04M1/656)} - H04M 3/5175: {Call or contact centers supervision arrangements} - H04M 2201/41: using speaker recognition - H04M 2203/6054: Biometric subscriber identification - H04M: TELEPHONIC COMMUNICATION (circuits for controlling other apparatus via a telephone cable and not involving telephone switching apparatus G08) - G10L: SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING ## Abstract Methods for improved fraudster detection in a call center. A subset of a plurality of voiceprints from a plurality of interactions between callers and agents at a call center can be used as the basis for fraudster detection. A plurality of connected components that represents one or more voiceprints can be determined based on the subset of the plurality of voiceprints. ## Claims ### Claim 1 A computerized method for improved fraudster detection in a call center, the method comprising: receiving, via a computing device, a plurality of voiceprints from a plurality of interactions between callers and agents at the call center; determining, via the computer, a subset of the plurality of voiceprints; determining, via the computer, for each of the plurality of voiceprints, a kth order moment score based on the subset of the plurality of voiceprints; sorting, via the computer, the plurality of voiceprints by their corresponding kth order moment score to obtain a sorted list of voiceprints; determining, via the computer, for each of the sorted plurality of voiceprints that are nearby in the sorted list, a clustered voiceprint pair for each pair of nearby voiceprints having a difference between their corresponding kth order moment score within a predetermined tolerance and adding the respective clustered voiceprint pair to a clustered voiceprint pair list; performing, by the computing device, a first search on the clustered voiceprint pair list to obtain a plurality of connected component voiceprints; determining, by the computing device, for each of the plurality of connected component voiceprints substantially in parallel, additional voiceprints that are filtered from the respective connected component voiceprint; and performing, by the computing device, a second search on the clustered voiceprint pair list to obtain a final plurality of connected component voiceprints, wherein each of the final plurality of connected components is an output that represents one or more voiceprints for use in fraud detection. ### Claim 2 (depends on 1) The computerized method of claim 1 wherein the kth order moment score is a mean score and a standard deviation score. ### Claim 3 (depends on 2) The computerized method of claim 2 wherein the sorting is based on the mean score. ### Claim 4 (depends on 2) The computerized method of claim 2 wherein voiceprints that are nearby are voiceprints that are within a fixed window, neighbors, or any combination thereof. ### Claim 5 (depends on 2) The computerized method of claim 2 wherein a clustered voiceprint pair has a difference between corresponding mean score within a mean score threshold. ### Claim 6 (depends on 2) The computerized method of claim 2 further comprising for each of the clustered voiceprint pairs in the clustered voiceprint pair list, if a difference between the corresponding standard deviation scores are above a predetermined threshold, removing, by the computing device, the respective clustered voiceprint pair. ### Claim 7 (depends on 1) The computerized method of claim 1 wherein the first search or the second search is a breadth-first-search, depth-first-search, or any combination thereof. ### Claim 8 (depends on 1) The computerized method of claim 1 wherein the subset of the plurality of voiceprints is based on an input value. ### Claim 9 (depends on 1) The computerized method of claim 1 wherein receiving the plurality of voiceprints further comprises recording calls received at the call center between the callers and agents in an audio file. ### Claim 10 A non-transitory computer program product comprising instructions which, when the program is executed cause the computer to: receive a plurality of voiceprints from a plurality of interactions between callers and agents at the call center; determine a subset of the plurality of voiceprints; determine for each of the plurality of voiceprints, a kth order moment score based on the subset of the plurality of voiceprints; sort the plurality of voiceprints by their corresponding kth order moment score to obtain a sorted list of voiceprints; determine for each of the sorted plurality of voiceprints that are nearby in the sorted list, a clustered voiceprint pair for each pair of nearby voiceprints having a difference between their corresponding kth order moment score within a predetermined tolerance and adding the respective clustered voiceprint pair to a clustered voiceprint pair list; perform a first search on the clustered voiceprint pair list to obtain a plurality of connected component voiceprints; determine for each of the plurality of connected component voiceprints substantially in parallel, additional voiceprints that are filtered from the respective connected component voiceprint; and perform a second search on the clustered voiceprint pair list to obtain a final plurality of connected component voiceprints, wherein each of the final plurality of connected components is an output that represents one or more voiceprints for use in fraud detection. ### Claim 11 (depends on 10) The computer program product of claim 10 wherein the kth order moment score is a mean score and a standard deviation score. ### Claim 12 (depends on 11) The computer program product of claim 11 wherein the sorting is based on the mean score. ### Claim 13 (depends on 11) The computer program product of claim 11 wherein voiceprints that are nearby are voiceprints that are within a fixed window, neighbors, or any combination thereof. ### Claim 14 (depends on 11) The computer program product of claim 11 wherein a clustered voiceprint pair has a difference between corresponding mean score within a mean score threshold. ### Claim 15 (depends on 11) The computer program product of claim 11 further comprising for each of the clustered voiceprint pairs in the clustered voiceprint pair list, if a difference between the corresponding standard deviation scores are above a predetermined threshold, removing, by the computing device, the respective clustered voiceprint pair. ### Claim 16 (depends on 10) The computer program product of claim 10 wherein the first search or the second search is a breadth-first-search, depth-first-search, or any combination thereof. ### Claim 17 (depends on 10) The computer program product of claim 10 wherein the subset of the plurality of voiceprints is based on an input value. ### Claim 18 (depends on 10) The computer program product of claim 10 wherein receiving the plurality of voiceprints further comprises recording calls received at the call center between the callers and agents in an audio file.