Adaptive confidence thresholds for speech recognition

1. A method for adjusting operation of a speech recognition engine implemented as a plurality of computer processes functioning in a computer processor and characterized by an associated receiver operating characteristic (ROC) curve, the method comprising:

in a first computer process, interpreting user confirmation of speech recognition results within a given confidence score threshold to create a confirmed portion of the ROC curve for the speech recognition engine;

in a second computer process, extending the confirmed portion of the ROC curve by extrapolation of unconfirmed speech recognition results beyond the confidence score threshold to generate an extended ROC curve; and

in a third computer process, adjusting the confidence score threshold based on the extended ROC curve to meet target operating constraints for operating the speech recognition engine to perform automatic speech recognition of user speech inputs.

2. A method according to claim 1 , wherein the confidence score threshold is an accept threshold such that speech recognition results having a confidence score below the accept threshold require user confirmation and speech recognition results having a confidence score above the accept threshold are accepted as correct without user confirmation.

3. A method according to claim 1 , wherein the confidence score threshold is a reject threshold such that speech recognition results having a confidence score above the reject threshold require user confirmation and speech recognition results having a confidence score below the reject threshold are rejected as incorrect without user confirmation.

4. A method according to claim 1 , wherein the extrapolation is based on a blind approach without user confirmation of speech recognition results beyond the confidence score threshold.

5. A method according to claim 1 , wherein the extrapolation is based on a sampling approach having user confirmation of selected speech recognition results beyond the confidence score threshold.

6. A speech recognition engine implemented as a plurality of computer processes functioning in a computer processor employing at least one hardware implemented computer processor and characterized by an associated receiver operating characteristic (ROC) curve, the engine comprising:

means for interpreting user confirmation of speech recognition results within a given confidence score threshold to create a confirmed portion of the ROC curve for the speech recognition engine;

means for extending the confirmed portion of the ROC curve by extrapolation of unconfirmed speech recognition results beyond the confidence score threshold to generate an extended ROC curve; and

means for adjusting the confidence score threshold based on the extended ROC curve to meet target operating constraints for operating the speech recognition engine to perform automatic speech recognition of user speech inputs.

7. An engine according to claim 6 , wherein the confidence score threshold is an accept threshold such that speech recognition results having a confidence score below the accept threshold require user confirmation and speech recognition results having a confidence score above the accept threshold are accepted as correct without user confirmation.

8. An engine according to claim 6 , wherein the confidence score threshold is a reject threshold such that speech recognition results having a confidence score above the reject threshold require user confirmation and speech recognition results having a confidence score below the reject threshold are rejected as incorrect without user confirmation.

9. An engine according to claim 6 , wherein the extrapolation is based on a blind approach without user confirmation of speech recognition results beyond the confidence score threshold.

10. An engine according to claim 6 , wherein the extrapolation is based on a sampling approach having user confirmation of selected speech recognition results beyond the confidence score threshold.

11. A computer program product implemented in a computer readable storage medium for adjusting operation of a speech recognition engine characterized by an associated receiver operating characteristic (ROC) curve, the product comprising:

program code for interpreting user confirmation of speech recognition results within a given confidence score threshold to create a confirmed portion of the ROC curve for the speech recognition engine;

program code for extending the confirmed portion of the ROC curve by extrapolation of unconfirmed speech recognition results beyond the confidence score threshold to generate an extended ROC curve; and

program code for adjusting the confidence score threshold based on the extended ROC curve to meet target operating constraints for operating the speech recognition engine to perform automatic speech recognition of user speech inputs.

12. A product according to claim 11 , wherein the confidence score threshold is an accept threshold such that speech recognition results having a confidence score below the accept threshold require user confirmation and speech recognition results having a confidence score above the accept threshold are accepted as correct without user confirmation.

13. A product according to claim 11 , wherein the confidence score threshold is a reject threshold such that speech recognition results having a confidence score above the reject threshold require user confirmation and speech recognition results having a confidence score below the reject threshold are rejected as incorrect without user confirmation.

14. A product according to claim 11 , wherein the extrapolation is based on a blind approach without user confirmation of speech recognition results beyond the confidence score threshold.

15. A product according to claim 11 , wherein the extrapolation is based on a sampling approach having user confirmation of selected speech recognition results beyond the confidence score threshold.

16. A speech recognition engine implemented as a plurality of computer processes functioning in a computer processor employing at least one hardware implemented computer processor and characterized by an associated receiver operating characteristic (ROC) curve, the engine comprising:

a results confirmation process that interprets user confirmation of speech recognition results within a given confidence score threshold to create a confirmed portion of the ROC curve for the speech recognition engine;

a curve extension process that extends the confirmed portion of the ROC curve by extrapolation of unconfirmed speech recognition results beyond the confidence score threshold to generate an extended ROC curve; and

a threshold adjustment process that adjusts the confidence score threshold based on the extended ROC curve to meet target operating constraints for operating the speech recognition engine to perform automatic speech recognition of user speech inputs.

17. An engine according to claim 16 , wherein the confidence score threshold is an accept threshold such that speech recognition results having a confidence score below the accept threshold require user confirmation and speech recognition results having a confidence score above the accept threshold are accepted as correct without user confirmation.

18. An engine according to claim 16 , wherein the confidence score threshold is a reject threshold such that speech recognition results having a confidence score above the reject threshold require user confirmation and speech recognition results having a confidence score below the reject threshold are rejected as incorrect without user confirmation.

19. An engine according to claim 16 , wherein the extrapolation is based on a blind approach without user confirmation of speech recognition results beyond the confidence score threshold.

20. An engine according to claim 16 , wherein the extrapolation is based on a sampling approach having user confirmation of selected speech recognition results beyond the confidence score threshold.