System and method for performing caller identity verification using multi-step voice analysis
Caller identity verification can be improved by employing a multi-step verification that leverages speech features that are obtained from multiple interactions with a caller. An enrollment is performed in which customer speech features and customer information are collected. When a caller calls into the call center, an attempt is made to verify the caller's identity by requesting the caller to speak a predefined phrase, extracting speech features from the spoken phrase, and comparing the phrase. If the purported identity of the caller can be matched with one of the customers based on the comparison, the identity of the caller is verified. If the match cannot be made with a high enough degree of confidence, the customer is asked to speak any phrase that is not predefined. Features are extracted from the caller's speech, combined with features previously extracted from the predefined speech, and compared to the enrollment features.
1. A system for performing caller identity verification using multi-step voice analysis with the aid of a digital computer, the digital computer comprising at least one computer processor configured to execute code to:
obtain one or more sets of stored speech features associated with a customer of a call center, wherein the one or more sets of stored speech features comprise speech features determined from at least one phrase provided by the customer during a voice connection with the call center;
receive a call from a caller into the call center;
identify a purported identity of the caller as the customer; and
verify the purported identity of the caller as the customer by:
prompting the caller to speak a text of the at least one phrase;
determining one or more speech features of the text-prompted speech;
prompting the caller to speak any additional phrase;
determining one or more speech features of the additional phrase;
combining the one or more speech features of the additional phrase with the one or more speech features of the text-prompted speech to form combined speech features;
comparing the combined speech features to one of the one or more sets of stored speech features; and
verifying the purported identity of the caller as the customer based on the comparison of the combined speech features to the one set of stored speech features.
2. The system according to claim 1 , wherein obtaining one or more sets of stored speech features comprises prompting the customer to speak while the customer is connected to the call center via the voice connection and wherein the processor when executing the code further:
identifies one or more characteristics of the voice connection associated with each of the one or more sets of stored speech features;
identifies one or more characteristics of the call;
compares the characteristics of the call and the characteristics for each of the one or more sets of stored speech features; and
identifies, based on the comparison of the characteristics of the call and the characteristics for each of the one or more sets of stored speech features, the one set of stored speech features used for the comparison with the combined speech features.
3. The system according to claim 2 , wherein the characteristics for the voice connection comprise a type of the voice connection and the characteristics for the call comprise a type of a connection between the caller and the call center during the call.
4. The system according to claim 2 , wherein the characteristics for the voice connection comprise a type of noise present during the voice connection and the characteristics for the call comprise a type of noise present during the call.
5. The system according to claim 1 , wherein the processor when executing the code:
performs speech recognition on the text-prompted speech; and
verifies that the text-prompted speech comprises the at least one phrase based on the speech recognition, wherein the purported identity is determined not to be the customer when the text-prompted speech is determined not to comprise the at least one phrase.
6. The system according to claim 1 , wherein the processor when executing the code:
receives a combination comprising a non-speech noise and at least one of the speech features of the text-prompted speech or the speech features of the additional phrase; and
filters the non-speech noise from the at least one of the speech features of the text-prompted speech or the speech features of the additional phrase in the combination.
7. The system according to claim 1 , wherein at least one of the speech features of the one or more sets of stored speech features, the speech features of the text-prompted speech, or the speech features of the additional phrase comprise at least one of a voice amplitude or a voice frequency.
8. The system according to claim 1 , wherein the processor when executing the code:
creates a user profile comprising the one or more sets of stored speech features and customer identifying information;
receives caller identifying information; and
compares the caller identifying information to the customer identifying information,
wherein the purported identity of the caller is further verified based on the comparison of the caller and customer identifying information.
9. The system according to claim 1 , wherein the processor when executing the code performs at least one of:
adding the combined speech features to the one or more sets of stored speech features; or
storing the combined speech features as an additional set of stored speech features.
10. The system according to claim 1 , wherein the customer and the caller are connected to the call center via at least one of a landline connection, a cellular network connection, or Internet telephony.
11. A method for performing caller identity verification using multi-step voice analysis with the aid of a digital computer, comprising:
obtaining, by at least one processor of the digital computer, one or more sets of stored speech features associated with a customer of a call center, wherein the one or more sets of stored speech features comprise speech features determined from at least one phrase;
receiving, by the at least one processor, a call from a caller into the call center;
identifying, by the at least one processor, a purported identity of the caller as the customer;
verifying, by the at least one processor, the purported identity of the caller as the customer by:
prompting the caller to speak a text of the at least one phrase;
determining one or more speech features of the text-prompted speech;
prompting the caller to speak any additional phrase;
determining one or more speech features of the additional phrase;
combining the one or more speech features of the additional phrase with the one or more speech features of the text-prompted speech to form combined speech features;
comparing the combined speech features to one of the one or more sets of stored speech features; and
verifying the purported identity of the caller as the customer based on the comparison of the combined speech features to the one set of stored speech features.
12. The method according to claim 11 , wherein obtaining one or more sets of stored speech features comprises prompting the customer to speak while the customer is connected to the call center via a voice connection, the method further comprising:
identifying, by the at least one processor, one or more characteristics of the voice connection associated with each of the one or more sets of stored speech features;
identifying, by the at least one processor, one or more characteristics of the call;
comparing, by the at least one processor, the characteristics of the call and the characteristics for each of the one or more sets of stored speech features; and
identifying, by the at least one processor, based on the comparison of the characteristics of the call and the characteristics for each of the one or more sets of stored speech features, the one set of stored speech features used for the comparison with the combined speech features.
13. The method according to claim 12 , wherein the characteristics for the voice connection comprise a type of the voice connection and the characteristics for the call comprise a type of a connection between the caller and the call center during the call.
14. The method according to claim 12 , wherein the characteristics for the voice connection comprise a type of noise present during the voice connection and the characteristics for the call comprise a type of noise present during the call.
15. The method according to claim 11 , further comprising:
performing, by the at least one processor, speech recognition on the text-prompted speech; and
verifying, by the at least one processor, that the text-prompted speech comprises the at least one phrase based on the speech recognition, wherein the purported identity is determined not to be the customer when the text-prompted speech is determined not to comprise the at least one phrase.
16. The method according to claim 11 , further comprising:
receiving, by the at least one processor, a combination comprising a non-speech noise and at least one of the speech features of the text-prompted speech or the speech features of the additional phrase; and
filtering, by the at least one processor, the non-speech noise from the at least one of the speech features of the text-prompted speech or the speech features of the additional phrase in the combination.
17. The method according to claim 11 , wherein at least one of the speech features of the one or more sets of stored speech features, the speech features of the text-prompted speech, or the speech features of the additional phrase comprise at least one of a voice amplitude or a voice frequency.
18. The method according to claim 11 , further comprising:
creating, by the at least one processor, a user profile comprising the one or more sets of stored speech features and customer identifying information;
receiving, by the at least one processor, caller identifying information; and
comparing, by the at least one processor, the caller identifying information to the customer identifying information,
wherein the purported identity of the caller is further verified based on the comparison of the caller and customer identifying information.
19. The method according to claim 11 , further comprising at least one of:
adding, by the at least one processor, the combined speech features to the one or more sets of stored speech features; or
storing, by the at least one processor, the combined speech features as an additional set of stored speech features.
20. The method according to claim 11 , wherein the customer and the caller are connected to the call center via at least one of a landline connection, a cellular network connection, or Internet telephony.