Application of machine learning techniques to select voice transformations

1. One or more non-transitory computer-readable media storing instructions, which when executed by one or more hardware processors, cause performance of operations comprising:

monitoring a conversation in real-time to detect one or more attributes of the conversation;

identifying a desired outcome of a conversation based on the one or more attributes;

identifying a plurality of voice modulations for accomplishing the desired outcome of the conversation;

determining a confidence score for each voice modulation of the plurality of voice modulations, the confidence score associated with a likelihood that the desired outcome will be accomplished subsequent to using the voice modulation to a participant's voice in the conversation, wherein determining the confidence score further comprises;

identifying at least one prior conversation having a prior desired outcome similar to the desired outcome of the conversation;

determining that one or more voice modulations of the plurality of voice modulations were used in the at least one prior conversation;

determining one or more rates at which the prior desired outcome of the at least one prior conversation was accomplished;

determining the confidence score of the one or more voice modulations in proportion to the rate at which the desired outcome were accomplished in the at least one prior conversation;

based at least on a first confidence score of accomplishing the desired outcome subsequent to using a first voice modulation of the plurality of voice modulations: selecting the first voice modulation from the plurality of voice modulations; and

applying the first voice modulation to the participant's voice.

2. The one or more media of claim 1 , wherein:

a conversation type is identified based on the one or more attributes of the conversation; and

the desired outcome of the conversation is based, at least in part, on the conversation type.

3. The one or more media of claim 2 , wherein:

the conversation type comprises one of a sales conversation, a help desk conversation, a customer retention conversation, and a customer care conversation; and

the desired outcome of the conversation comprises completing a sale, answering a question, and resolving a customer complaint, respectively.

4. The one or more media of claim 1 , wherein the one or more voice modulations applied to the voice of the participant comprises one or more of altering a tone, a speed, and a regional accent.

5. The one or more media of claim 1 , wherein the one or more voice modulations applied to the voice of the participant comprises one or more of altering a content density.

6. The one or more media of claim 1 , wherein determining the confidence score further comprises:

identifying a set of voice modulations used in a prior conversation having a desired outcome similar to the desired outcome for the conversation;

determining similarity scores between the set of voice modulations used in the prior conversation and corresponding voice modulations of the one or more voice modulations; and

using the similarity scores as a factor in determining the confidence scores of the corresponding one or more voice modulations.

7. The one or more media of claim 1 , wherein determining the confidence score further comprises basing the confidence score on a rate at which desired outcomes are accomplished.

8. A method comprising:

monitoring a conversation in real-time to detect one or more attributes of the conversation;

identifying a desired outcome of a conversation based on the one or more attributes;

identifying a plurality of voice modulations for accomplishing the desired outcome of the conversation;

determining a confidence score for each voice modulation of the plurality of voice modulations, the confidence score associated with a likelihood that the desired outcome will be accomplished subsequent to using the voice modulation to a participant's voice in the conversation, wherein determining the confidence score further comprises;

identifying at least one prior conversation having a prior desired outcome similar to the desired outcome of the conversation;

determining that one or more voice modulations of the plurality of voice modulations were used in the at least one prior conversation;

determining one or more rates at which the prior desired outcome of the at least one prior conversation was accomplished;

determining the confidence score of the one or more voice modulations in proportion to the rate at which the desired outcome were accomplished in the at least one prior conversation;

based at least on a first confidence score of accomplishing the desired outcome subsequent to using a first voice modulation of the plurality of voice modulations: selecting the first voice modulation from the plurality of voice modulations; and

applying the first voice modulation to the participant's voice.

9. The method of claim 8 , wherein:

a conversation type is identified based on the one or more attributes of the conversation; and

the desired outcome of the conversation is based, at least in part, on the conversation type.

10. The method of claim 9 , wherein:

the conversation type comprises one of a sales conversation, a help desk conversation, a customer retention conversation, and a customer care conversation; and

the desired outcome of the conversation comprises completing a sale, answering a question, and resolving a customer complaint, respectively.

11. The method of claim 8 , wherein the one or more voice modulations applied to the voice of the participant comprises one or more of altering a tone, a speed, and a regional accent.

12. The method of claim 8 , wherein the one or more voice modulations applied to the voice of the participant comprises one or more of altering a content density.

13. The method of claim 8 , wherein determining the confidence score further comprises:

identifying a set of voice modulations used in a prior conversation having a desired outcome similar to the desired outcome for the conversation;

determining similarity scores between the set of voice modulations used in the prior conversation and corresponding voice modulations of the one or more voice modulations; and

using the similarity scores as a factor in determining the confidence scores of the corresponding one or more voice modulations.

14. The method of claim 8 , wherein determining the confidence score further comprises basing the confidence score on a rate at which desired outcomes are accomplished.

15. A system comprising:

at least one device including a hardware processor;

the system being configured to perform operations comprising:

monitoring a conversation in real-time to detect one or more attributes of the conversation;

identifying a desired outcome of a conversation based on the one or more attributes;

identifying a plurality of voice modulations for accomplishing the desired outcome of the conversation;

determining a confidence score for each voice modulation of the plurality of voice modulations, the confidence score associated with a likelihood that the desired outcome will be accomplished subsequent to using the voice modulation to a participant's voice in the conversation, wherein determining the confidence score further comprises;

identifying at least one prior conversation having a prior desired outcome similar to the desired outcome of the conversation;

determining that one or more voice modulations of the plurality of voice modulations were used in the at least one prior conversation;

determining one or more rates at which the prior desired outcome of the at least one prior conversation was accomplished;

determining the confidence score of the one or more voice modulations in proportion to the rate at which the desired outcome were accomplished in the at least one prior conversation;

based at least on a first confidence score of accomplishing the desired outcome subsequent to using a first voice modulation of the plurality of voice modulations:

selecting the first voice modulation from the plurality of voice modulations; and

applying the first voice modulation to the participant's voice.

16. The system of claim 15 , wherein:

a conversation type is identified based on the one or more attributes of the conversation; and

the desired outcome of the conversation is based, at least in part, on the conversation type.

17. The system of claim 16 , wherein:

the conversation type comprises one of a sales conversation, a help desk conversation, a customer retention conversation, and a customer care conversation; and

the desired outcome of the conversation comprises completing a sale, answering a question, and resolving a customer complaint, respectively.

18. The system of claim 15 , wherein the one or more voice modulations applied to the voice of the participant comprises one or more of altering a tone, a speed, and a regional accent.

19. The system of claim 15 , wherein the one or more voice modulations applied to the voice of the participant comprises one or more of altering a content density.

20. The system of claim 15 , wherein determining the confidence score further comprises:

identifying a set of voice modulations used in a prior conversation having a desired outcome similar to the desired outcome for the conversation;

determining similarity scores between the set of voice modulations used in the prior conversation and corresponding voice modulations of the one or more voice modulations; and

using the similarity scores as a factor in determining the confidence scores of the corresponding one or more voice modulations.

21. The system of claim 15 , wherein determining the confidence score further comprises basing the confidence score on a rate at which desired outcomes are accomplished.