Systems, methods, and media for determining fraud risk from audio signals

Abstract

Systems, methods, and media for determining fraud risk from audio signals and non-audio data are provided herein. Some exemplary methods include receiving an audio signal and an associated audio signal identifier, receiving a fraud event identifier associated with a fraud event, determining a speaker model based on the received audio signal, determining a channel model based on a path of the received audio signal, using a server system, updating a fraudster channel database to include the determined channel model based on a comparison of the audio signal identifier and the fraud event identified, and updating a fraudster voice database to include the determined speaker model based on a comparison of the audio signal identifier and the fraud event identifier.

1 . A method for screening callers in a call center, the method comprising:
 maintaining a set of channel models, wherein each channel model represents noise, artifacts, distortions, degradations, or modifications of call audio data associated with an instance of fraud;   receiving a screening request, the screening request comprising an audio sample from a caller;   extracting a caller's channel model from the audio sample;   comparing the caller's channel model with channel models in the set of channel models;   generating a channel match score based on the comparison, the channel match score indicating the level of risk that the caller is a fraudster; and   providing the channel match score to a call agent that is currently speaking with the caller.   
     
     
         2 . The method according to  claim 1 , wherein the comparing the caller's speaker model with channel models in the set of channel models, comprises:
 selecting channel models from the set of channel models based on past calls associated with the caller's account; and   comparing the caller's channel model with the selected channel models.   
     
     
         3 . The method according to  claim 1 , further comprising:
 maintaining a set of speaker models, wherein each speaker model represents voice characteristics and linguistic characteristics of a fraudster;   extracting a caller's speaker model from the audio sample;   comparing the caller's speaker model with speaker models in the set of speaker models;   generating a voice match score based on the comparison, the voice match score indicating the level of risk that the caller is a fraudster; and   providing the voice match score to the call agent that is currently speaking with the caller.   
     
     
         4 . The method according to  claim 3 , wherein the comparing the caller's speaker model with speaker models in the set of speaker models, comprises:
 selecting, based on the channel match score, speaker models from the set of speaker models; and   comparing the caller's speaker model with the selected speaker models.   
     
     
         5 . The method according to  claim 3 , further comprising:
 generating a risk score based on the voice match score and the channel match score; and   providing the risk score to the call agent that is currently speaking with the caller, the risk match score indicating the level of risk that the caller is a fraudster.   
     
     
         6 . The method according to  claim 1 , wherein the caller's channel model comprises noise, artifacts, distortions, degradations, or modifications resulting from a telecommunications path between the caller and the call center. 
     
     
         7 . The method according to  claim 6 , wherein the noise, artifacts, distortions, degradations, or modifications indicate a landline, VoIP phone, or cellular phone. 
     
     
         8 . The method according to  claim 6 , wherein the noise, artifacts, distortions, degradations, or modifications indicate a CDMA, GSM, or VOIP communication method. 
     
     
         9 . The method according to  claim 6 , wherein the noise, artifacts, distortions, degradations, or modifications indicate a geographic region of the caller. 
     
     
         10 . The method according to  claim 1 , wherein the caller's channel model comprises noise, artifacts, distortions, degradations, or modifications resulting from one or more devices used by the caller. 
     
     
         11 . The method according to  claim 10 , wherein the one or more devices comprises a voice changer. 
     
     
         12 . The method according to  claim 10 , wherein the one or more devices comprises a microphone in a handset used by the caller. 
     
     
         13 . A system for screening callers in a call center, the system comprising:
 a call database storing a set of channel models, wherein each channel model represents noise, artifacts, distortions, degradations, or modifications of call audio data associated with an instance of fraud; and   a computing system in communication with the call database comprising a processor, a display system, and a memory, wherein the memory stores computer-readable instructions causing the processor to perform operations comprising:
 receiving a screening request, the screening request comprising an audio sample from a caller, 
 extracting a caller's channel model from the audio sample, 
 comparing the caller's channel model with channel models in the set of channel models, 
 generating a channel match score based the comparison, the channel match score indicating the level of risk that the caller is a fraudster, and 
 transmitting the channel match score to the display system used by a call agent that is currently speaking with the caller. 
   
     
     
         14 . The system according to  claim 13 , wherein the caller's channel model comprises noise, artifacts, distortions, degradations, or modifications resulting from a telecommunications path between the caller and the call center. 
     
     
         15 . The system according to  claim 14 , wherein the noise, artifacts, distortions, degradations, or modifications indicate a landline, VoIP phone, or cellular phone. 
     
     
         16 . The system according to  claim 14 , wherein the noise, artifacts, distortions, degradations, or modifications indicate a CDMA, GSM, or VOIP communication type. 
     
     
         17 . The system according to  claim 14 , wherein the noise, artifacts, distortions, degradations, or modifications indicate a geographic region of the caller. 
     
     
         18 . The system according to  claim 13 , wherein the caller's channel model comprises noise, artifacts, distortions, degradations, or modifications resulting from one or more devices used by the caller. 
     
     
         19 . The system according to  claim 18 , wherein the one or more devices comprises a voice changer. 
     
     
         20 . A non-transitory tangible computer readable storage medium containing computer readable program code that when executed by a processor of a computing device cause the computing device to perform operations comprising:
 maintaining a set of channel models, wherein each channel model represents noise, artifacts, distortions, degradations, or modifications of call audio data associated with an instance of fraud;   receiving a screening request, the screening request comprising an audio sample from a caller;   extracting a caller's channel model from the audio sample;   comparing the caller's channel model with channel models in the set of channel models;   generating a channel match score based on the comparison, the channel match score indicating the level of risk that the caller is a fraudster; and   providing the channel match score to a call agent that is currently speaking with the caller.