Intelligent personalized speech recognition

1. A video game system comprising:

an audio input interface configured to receive audio data from one or more audio input devices;

a display output configured for transmitting graphics data;

a data store including computer-executable instructions; and

a processor configured to execute the computer-executable instructions to perform a method comprising:

determining a first phoneme mapping for each letter or combination of letters in a grammar training set;

receiving audio data comprising a speech sample of a user reading a grammar training set;

recognizing the speech sample using the first phoneme mapping;

determining a first confidence score indicative of how accurately the grammar training set is recognized using the speech sample using the first phoneme mapping;

changing the first phoneme mapping to generate a mutated phoneme mapping until a confidence score generated using the mutated phoneme mapping satisfies a defined confidence threshold;

storing, in a user profile in the data store, the mutated phoneme mapping that satisfies the defined confidence threshold, wherein the mutated phoneme mapping is associated with the user in the user profile;

receiving subsequent audio data from the user;

accessing the user profile to retrieve the mutated phoneme mapping that is associated with the user; and

using the mutated phoneme mapping to recognize the subsequent audio data received from the user.

2. The video game system of claim 1 , wherein the processor is configured to execute the computer-executable instructions to perform the method further comprising:

receiving information indicative of an accent of the user; and

selecting the first phoneme mapping based on the information indicative of an accent.

3. The video game system of claim 2 , wherein the information indicative of an accent includes at least one of:

an age of the user;

a gender of the user;

a current location of the user; or

a location where the user lived.

4. The video game system of claim 1 , wherein the processor is configured to execute the computer-executable instructions to perform the method further comprising:

changing a phoneme associated with one or more letters in the grammar training set;

recognizing the speech sample using the mutated phoneme mapping;

determining a second confidence score based on an analysis of the speech sample using the mutated phoneme mapping;

comparing the second confidence score to the first confidence score; and

based at least in part on the comparison of the second confidence score to the first confidence score, undoing a previous mutation and performing a different mutation to generate the mutated phoneme mapping.

5. The video game system of claim 1 , wherein changing the first phoneme mapping to generate the mutated phoneme mapping until the confidence score generated using the mutated phoneme mapping satisfies the defined confidence threshold comprises:

changing a phoneme associated with one or more letters in the grammar training set;

recognizing the speech sample using the mutated phoneme mapping;

determining a second confidence score based on an analysis of the speech sample using the mutated phoneme mapping;

comparing the second confidence score to the first confidence score; and

determining that the grammar training set is more accurately recognized using the mutated phoneme mapping as compared to using the first phoneme mapping.

6. The video game system of claim 1 , wherein changing the first phoneme mapping comprises randomly changing one or more phoneme mappings.

7. The video game system of claim 1 , wherein the first confidence score is a per phoneme confidence score.

8. A method for voice recognition as implemented by a computing system configured with specific computer-executable instructions, the method comprising:

receiving a grammar training set;

receiving a speech sample of a user reading at least a first portion of the grammar training set;

determining a first phoneme mapping for each letter or combination of letters in the first portion of the grammar training set;

recognizing the speech sample using the first phoneme mapping;

determining a first confidence score indicative of how accurately the recognized speech sample corresponds to the first portion of the grammar training set by using the first phoneme mapping;

changing the first phoneme mapping to generate a mutated phoneme mapping until a confidence score generated using the mutated phoneme mapping satisfies a defined confidence threshold;

storing, in a user profile in a data store, the mutated phoneme mapping that satisfies the defined confidence threshold, wherein the mutated phoneme mapping is associated with the user in the user profile;

receiving subsequent audio data from the user;

accessing the user profile to retrieve the mutated phoneme mapping that is associated with the user; and

using the mutated phoneme mapping to recognize the subsequent audio data received from the user.

9. The method of claim 8 , further comprising:

receiving information indicative of an accent of the user; and

selecting the first phoneme mapping based on the information indicative of an accent.

10. The method of claim 9 , wherein the information indicative of an accent includes at least one of:

an age of the user;

a gender of the user;

a current location of the user; or

a location where the user lived.

11. The method of claim 8 , further comprising:

changing a phoneme associated with one or more letters in the grammar training set;

recognizing the speech sample using the mutated phoneme mapping;

determining a second confidence score based on an analysis of the speech sample using the mutated phoneme mapping;

comparing the second confidence score to the first confidence score; and

based at least in part on the comparison of the second confidence score to the first confidence score, performing a different mutation to generate the mutated phoneme mapping.

12. The method of claim 8 , wherein changing the first phoneme mapping to generate the mutated phoneme mapping until the confidence score generated using the mutated phoneme mapping satisfies the defined confidence threshold comprises:

changing a phoneme associated with one or more letters in the grammar training set;

recognizing the speech sample using the mutated phoneme mapping;

determining a second confidence score based on an analysis of the speech sample using the mutated phoneme mapping;

comparing the second confidence score to the first confidence score; and

determining that the grammar training set is more accurately recognized using the mutated phoneme mapping as compared to using the first phoneme mapping.

13. The method of claim 8 , wherein changing the first phoneme mapping comprises randomly changing one or more phoneme mappings.

14. The method of claim 8 , further comprising:

detecting a word in the subsequent audio data, wherein the word is not included in the grammar training set.

15. A non-transitory, computer-readable storage medium storing computer readable instructions that, when executed by one or more processors in a computing device, causes the computing device to perform operations comprising:

receiving, through a network connection port, a speech sample of a user reading a grammar training set;

determining a first phoneme mapping for each letter or combination of letters in the grammar training set;

recognizing the speech sample using the first phoneme mapping;

determining a first confidence score indicative of how accurately the grammar training set is recognized using the speech sample using the first phoneme mapping;

changing the first phoneme mapping to generate a mutated phoneme mapping until a confidence score generated using the mutated phoneme mapping satisfies a defined confidence threshold;

storing, in a user profile in a data store, the mutated phoneme mapping that satisfies the defined confidence threshold, wherein the mutated phoneme mapping is associated with the user in the user profile;

receiving subsequent audio data from the user;

accessing the user profile to retrieve the mutated phoneme mapping that is associated with the user; and

using the mutated phoneme mapping to recognize the subsequent audio data received from the user.

16. The computer-readable storage medium of claim 15 , wherein the computer readable instructions are further configured to cause the computing device to perform operations comprising:

receiving information indicative of an accent of the user; and

selecting the first phoneme mapping based on the information indicative of an accent.

17. The computer-readable storage medium of claim 16 , wherein the information indicative of an accent includes at least one of:

an age of the user;

a gender of the user;

a current location of the user; or

a location where the user lived.

18. The computer-readable storage medium of claim 15 , wherein the computer readable instructions are further configured to cause the computing device to perform operations comprising:

changing a phoneme associated with one or more letters in the grammar training set;

recognizing the speech sample using the mutated phoneme mapping;

determining a second confidence score based on an analysis of the speech sample using the mutated phoneme mapping;

comparing the second confidence score to the first confidence score; and

based at least in part on the comparison of the second confidence score to the first confidence score, undo a previous mutation and performing a different mutation to generate the mutated phoneme mapping.

19. The computer-readable storage medium of claim 15 , wherein changing the first phoneme mapping to generate the mutated phoneme mapping until the confidence score generated using the mutated phoneme mapping satisfies the defined confidence threshold comprises:

changing a phoneme associated with one or more letters in the grammar training set;

recognizing the speech sample using the mutated phoneme mapping;

determining a second confidence score based on an analysis of the speech sample using the mutated phoneme mapping;

comparing the second confidence score to the first confidence score; and

determining that the grammar training set is more accurately recognized using the mutated phoneme mapping as compared to using the first phoneme mapping.

20. The computer-readable storage medium of claim 15 , wherein the first confidence score is a per word confidence score.