Voice synthesis for virtual agents

1. A computer-implemented method performed by at least one processor, the method comprising:

receiving information identifying a customer and an intent of the customer;

determining, based on the information identifying the customer, if the customer is a known customer;

in response to determining that the customer is an unknown customer, selecting a first voice for a virtual agent and assigning the first voice to the virtual agent during communications with the customer, the first voice including at least one first sound property;

in response to determining that the customer is a known customer, selecting a second voice for the virtual agent, including selecting at least one second sound property for a voice that is an alteration of the at least one first sound property, based on previously received information about the customer, and assigning the second voice to the virtual agent during communications with the customer; and

determining, based on the intent of the customer, if the voice assigned to the virtual agent should be altered, and in response to determining that the voice assigned to the virtual agent should be altered, altering the at least one first sound property or the at least one second sound property.

2. The computer-implemented method of claim 1 , further comprising:

monitoring the communications with the customer; and

in response to detecting a second intent of the customer during the monitoring, altering the at least one first sound property or the at least one second sound property based on the detected change in intent.

3. The computer-implemented method of claim 1 , further comprising:

performing signal processing on received voice data to identify at least one sound property of a voice of the customer, wherein selecting the first voice for the virtual agent is further based on the at least one sound property of the voice of the customer.

4. The computer-implemented method of claim 1 , wherein:

assigning the first voice to the virtual agent includes synthesizing the first voice using speech synthesis;

assigning the second voice to the virtual agent includes synthesizing the second voice using speech synthesis; and

the virtual agent uses at least one of the synthesized first voice and the synthesized second voice during communications with the customer.

5. The computer-implemented method of claim 1 , further comprising:

receiving textual message information from a customer service representative to the customer; and

altering the voice assigned to the virtual agent based on the textual message information.

6. The computer-implemented method of claim 1 , wherein selecting the second voice includes matching the customer to a voice profile, the voice profile identified by:

identifying, using one or more machine learning techniques, one or more previous communications between the customer and a human service representative that resulted in a positive outcome; and

performing signal processing on voice data representing a voice of the human service representative.

7. The computer-implemented method of claim 1 , wherein the intent of the customer is determined by performing signal processing voice data indicating a voice of the customer.

8. The computer-implemented method of claim 1 , further comprising:

monitoring the communications with the customer;

determining a response of the customer to the voice assigned to the virtual agent; and

updating a machine learning system based on the determined response.

9. The computer-implemented method of claim 1 , wherein the first voice or the second voice is selected using a machine-learning system to match the voice of a service representative assigned to the customer and is used in real-time in two-way communications with the customer.

10. A system, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor, the memory storing instructions which, when executed by the at least one processor, cause the at least one processor to perform operations comprising:

receiving information identifying a customer and an intent of the customer;

determining, based on the information identifying the customer, if the customer is a known customer;

in response to determining that the customer is an unknown customer, selecting a first voice for a virtual agent and assigning the first voice to the virtual agent during communications with the customer, the first voice including at least one first sound property;

in response to determining that the customer is a known customer, selecting a second voice for the virtual agent, including selecting at least one second sound property for a voice that is an alteration of the at least one first sound property, based on previously received information about the customer, and assigning the second voice to the virtual agent during communications with the customer; and

determining, based on the intent of the customer, if the voice assigned to the virtual agent should be altered, and in response to determining that the voice assigned to the virtual agent should be altered, altering the at least one first sound property or the at least one second sound property.

11. The system of claim 10 , wherein the operations further comprise:

monitoring the communications with the customer; and

in response to detecting second intent of the customer during the monitoring, altering the at least one first sound property or the at least one second sound property based on the detected change in intent.

12. The system of claim 10 , wherein the operations further comprise:

performing signal processing on received voice data to identify at least one sound property of a voice of the customer, wherein selecting the first voice for the virtual agent is further based on the at least one sound property of the voice of the customer.

13. The system of claim 10 , wherein:

assigning the first voice to the virtual agent includes synthesizing the first voice using speech synthesis;

assigning the second voice to the virtual agent includes synthesizing the second voice using speech synthesis; and

the virtual agent uses at least one of the synthesized first voice and the synthesized second voice during communications with the customer.

14. The system of claim 10 , wherein the operations further comprise:

receiving textual message information from a customer service representative to the customer; and

altering the voice assigned to the virtual agent based on the textual message information.

15. The system of claim 10 , wherein selecting the second voice includes matching the customer to a voice profile, the voice profile identified by:

identifying, using one or more machine learning techniques, one or more previous communications between the customer and a human service representative that resulted in a positive outcome; and

performing signal processing on voice data representing a voice of the human service representative.

16. The system of claim 10 , wherein the intent of the customer is determined by performing signal processing voice data indicating a voice of the customer.

17. The system of claim 10 , wherein the operations further comprise:

monitoring the communications with the customer;

determining a response of the customer to the voice assigned to the virtual agent and updating a machine learning system based on the determined response.

18. The system of claim 10 , wherein the first voice or the second voice is selected using a machine-learning system to match the voice of a service representative assigned to the customer for the duration of a call and is used in real-time in two-way communications with the customer.

19. One or more computer-readable media storing instructions which, when executed by at least one processor, cause the at least one processor to perform operations comprising:

receiving information identifying a customer and an intent of the customer;

determining, based on the information identifying the customer, if the customer is a known customer;

in response to determining that the customer is an unknown customer, selecting a first voice for a virtual agent and assigning the first voice to the virtual agent during communications with the customer, the first voice including at least one first sound property;

in response to determining that the customer is a known customer, selecting a second voice for the virtual agent, including selecting at least one second sound property for a voice that is an alteration of the at least one first sound property, based on previously received information about the customer, and assigning the second voice to the virtual agent during communications with the customer; and

determining, based on the intent of the customer, if the voice assigned to the virtual agent should be altered, and in response to determining that the voice assigned to the virtual agent should be altered, altering the at least one first sound property or the at least one second sound property.

20. The one or more computer-readable media of claim 19 , wherein the operations further comprise:

monitoring the communications with the customer; and

in response to detecting a change in second intent of the customer during the monitoring, altering the at least one first sound property or the at least one second sound property based on the detected change in intent.

21. The one or more computer-readable media of claim 19 , wherein the operations further comprise:

performing signal processing on received voice data to identify at least one sound property of a voice of the customer, wherein selecting the first voice for the virtual agent is further based on the at least one sound property of the voice of the customer.

22. The one or more computer-readable media of claim 19 , wherein

assigning the first voice to the virtual agent includes synthesizing the first voice using speech synthesis;

assigning the second voice to the virtual agent includes synthesizing the second voice using speech synthesis; and

the virtual agent uses at least one of the synthesized first voice and the synthesized second voice during communications with the customer.

23. The one or more computer-readable media of claim 19 , wherein the operations further comprise:

receiving textual message information from a customer service representative to the customer; and

altering the voice assigned to the virtual agent based on that the textual message information.

24. The one or more computer-readable media of claim 19 , wherein selecting the second voice includes matching the customer to a voice profile, the voice profile identified by:

identifying, using one or more machine learning techniques, one or more previous communications between the customer and a human service representative that resulted in a positive outcome; and

performing signal processing on voice data representing a voice of the human service representative.

25. The one or more computer-readable media of claim 19 , wherein the intent of the customer is determined by performing signal processing voice data indicating a voice of the customer.

26. The one or more computer-readable media of claim 19 , wherein the operations further comprise:

monitoring the communications with the customer;

determining a response of the customer to the voice assigned to the virtual agent; and

updating a machine learning system based on the determined response.

27. The one or more computer-readable media of claim 19 , The computer-implemented method of claim 1 , wherein the first voice or the second voice is selected using a machine-learning system to match the voice of a service representative assigned to the customer and is used in real-time in two-way communications with the customer.