Method and system for dynamic nametag scoring

• US 20070174055A1
• Filed: 01/20/2006
• Published: 07/26/2007
• Est. Priority Date: 01/20/2006
• Status: Active Grant

• Current Assignee: General Motors LLC (General Motors Company)
• Original Assignee: General Motors Corporation (Motors Liquidation Co. GUC Trust)
• Inventors: Chengalvarayan, Rathinavelu, Correia, John
• Granted Patent: US 8,626,506 B2
• Time in Patent Office: Days
• US Class Current: 704/251
• CPC Class Codes: G10L 15/08 Speech classification or search

Abstract

A method for dynamic nametag scoring includes receiving at least one confusion table including at least one circumstantial condition wherein the confusion table is based on a plurality of phonetically balanced utterances, determining a plurality of templates for the nametag based on the received confusion tables, and determining a global nametag score for the nametag based on the determined templates. A computer usable medium with suitable computer program code is employed for dynamic nametag scoring.

Claims

1. A method for dynamic nametag scoring comprising:

receiving at least one confusion table including at least one circumstantial condition wherein the confusion table is based on a plurality of phonetically balanced utterances;

determining a plurality of templates for the nametag based on the received confusion tables; and

determining a global nametag score for the nametag based on the determined templates.

2. The method of claim 1 wherein determining the plurality of templates for at least one nametag comprises:

receiving a nametag utterance at a speech recognition engine;

determining at least one best template for the received nametag utterance at the speech recognition engine based on a stored model; and

modifying the best template based on at least one confusion matrix to derive a plurality of templates corresponding to the best template.

3. The method of claim 1 wherein determining the global nametag score comprises:

determining at least one template score for each template;

determining at least one weight for each template;

combining the template score and the weight to yield a weighted template score; and

calculating a linear combination of the weighted template scores for all templates associated with a particular nametag.

4. The method of claim 3 wherein determining at least one weight for each template comprises:

determining a nametag selection count;

comparing the nametag selection count to a threshold count;

determining a template selection count; and

calculating the quotient of the template selection count and the total selection count.

5. The method of claim 4 further comprising:

modifying the template selection count based on a nametag retrieval.

6. The method of claim 1 further comprising:

continuously updating the global nametag score based on an updated weight determination.

7. The method of claim 1 further comprising:

selecting the templates based on at least one selection factor.

8. The method of claim 7 wherein selecting the templates based on at least one selection factor comprises:

comparing the plurality of templates to a nametag audio string based on an acoustic model;

determining at least one selection factor based on the comparison;

ranking the plurality of templates based on the selection factors; and

deleting at least one template based on the ranking.

9. The method of claim 1 wherein the circumstantial condition is a user gender, a user dialect, a vehicle type, a vehicle operating condition, or a noise condition.

10. A computer usable medium storing a computer program including computer program code for dynamic nametag scoring, the computer usable medium comprising:

computer program code for receiving at least one confusion table including one circumstantial condition wherein the confusion table is based on a plurality of phonetically balanced utterances;

computer program code for determining a plurality of templates for the nametag based on the received confusion tables; and

computer program code for determining a global nametag score for the nametag based on the determined templates.

11. The computer usable medium of claim 10 wherein computer program code for determining the plurality of templates for at least one nametag comprises:

computer program code for receiving a nametag utterance at a speech recognition engine;

computer program code for determining at least one best template for the received nametag utterance at the speech recognition engine based on a stored model; and

computer program code for modifying the best template based on at least one confusion matrix to derive a plurality of templates corresponding to the best template.

12. The method computer usable medium of claim 10 wherein computer program code for determining the global nametag score comprises:

computer program code for determining at least one template score for each template;

computer program code for determining at least one weight for each template;

computer program code for combining the template score and the weight to yield a weighted template score; and

computer program code for calculating a linear combination of the weighted template scores for all templates associated with a particular nametag.

13. The computer usable medium of claim 12 wherein computer program code for determining a plurality of weights comprises:

computer program code for determining a nametag selection count;

computer program code for comparing the nametag selection count to a threshold count;

computer program code for determining a template selection count; and

computer program code for calculating the quotient of the template selection count and the total selection count.

14. The computer usable medium of claim 13 further comprising:

computer program code for modifying the template selection count based on a nametag retrieval.

15. The computer usable medium of claim 10 further comprising:

computer program code for continuously updating the global nametag score based on an updated weight determination.

16. The computer usable medium of claim 10 further comprising:

computer program code for selecting the templates based on at least one selection factor.

17. The computer usable medium of claim 16 wherein computer program code for selecting the templates comprises:

computer program code for comparing a plurality of nametag utterances to the plurality of templates based on an acoustic model;

computer program code for determining at least one selection factor based on the comparison;

computer program code for ranking the plurality of templates based on the selection factors; and

computer program code for deleting at least one template based on the comparison.

18. A system for dynamic nametag scoring comprising:

means for receiving at least one confusion table including at least one circumstantial condition wherein the confusion table is based on a plurality of phonetically balanced utterances;

means for determining a plurality of templates for the nametag based on the confusion tables; and

means for determining a global nametag score for the nametag based on the determined templates.

19. A method for determining a plurality of confusion tables comprising:

receiving a plurality of test utterances;

generating a plurality of component sequences based on the test utterances and at least one circumstantial condition;

determining a confusion matrix based on the component sequences and the test utterances; and

generating a confusion table based on the confusion matrix.