Method and system for identifying and correcting accent-induced speech recognition difficulties

• US 20060020463A1
• Filed: 07/22/2004
• Published: 01/26/2006
• Est. Priority Date: 07/22/2004
• Status: Active Grant

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: International Business Machines Corporation
• Inventors: Reich, David E.
• Granted Patent: US 8,036,893 B2
• Time in Patent Office: Days
• US Class Current: 704/257
• CPC Class Codes: G10L 15/187 Phonemic context, e.g. pronunciation rules, phonotactical constraints or phoneme n-grams
• CPC Class Codes: G10L 2015/025 Phonemes, fenemes or fenones being the recognition units
• CPC Class Codes: G10L 2015/227 of the speaker; Human-factor methodology

Abstract

A system for use in speech recognition includes an acoustic module accessing a plurality of distinct-language acoustic models, each based upon a different language; a lexicon module accessing at least one lexicon model; and a speech recognition output module. The speech recognition output module generates a first speech recognition output using a first model combination that combines one of the plurality of distinct-language acoustic models with the at least one lexicon model. In response to a threshold determination, the speech recognition output module generates a second speech recognition output using a second model combination that combines a different one of the plurality of distinct-language acoustic models with the at least one distinct-language lexicon model.

Claims

1. A system for use in speech recognition, the system comprising:

an acoustic module accessing a plurality of distinct-language acoustic models, each based upon a different language;

a lexicon module accessing at least one lexicon model; and

a speech recognition output module for

generating a first speech recognition output using a first model combination that combines one of the plurality of distinct-language acoustic models with the at least one lexicon model, and

in response to a threshold determination, generating a second speech recognition output using a second model combination that combines a different one of the plurality of distinct-language acoustic models with the at least one distinct-language lexicon model.

2. The system as defined in claim 1, further comprising a threshold determination module for making the threshold determination by comparing a first confidence score for the first speech recognition output with a minimum acceptable score.

3. The system as defined in claim 1, further comprising a confidence score generation module generating confidence scores for each speech recognition output.

4. The system as defined in claim 3, further comprising a best model combination determination module determining a best model combination based upon the confidence scores.

5. The system as defined in claim 1, wherein the at least one lexicon model comprises a plurality of distinct-language lexicon models, each based upon a different language.

6. The system as defined in claim 5, wherein the speech recognition output module generates additional speech recognition outputs using different model combinations comprising different pairings of the plurality distinct-language acoustic models with the plurality of distinct language lexicon models.

7. The system as defined in claim 6, further comprising a confidence score generation module generating confidence scores for each speech recognition output.

8. The system as defined in claim 7, further comprising a best model combination determination module determining a best model combination based upon the confidence scores.

9. A method of generating speech recognition output using a plurality of distinct-language acoustic models, each based upon a different language, and at least one lexicon model, the method comprising the steps of:

generating a first speech recognition output using a first model combination that combines one of the plurality of distinct-language acoustic models with the at least one lexicon model; and

in response to a threshold determination, generating a second speech recognition output using a second model combination that combines a different one of the plurality of distinct-language acoustic models with the at least one distinct-language lexicon model.

10. The method as defined in claim 9, further comprising determining a first confidence score for the first speech recognition output, and wherein the threshold determination is based upon a comparison of the first confidence score with a minimum acceptable score.

11. The method of claim 9, further comprising generating confidence scores for each speech recognition output and determining a best model combination based upon the confidence scores.

12. The method as defined in claim 9, wherein the at least one lexicon model comprises a plurality of distinct-language lexicon models, each based upon a different language.

13. The method as defined in claim 12, further comprising generating additional speech recognition outputs using different model combinations comprising different pairings of the plurality distinct-language acoustic models with the plurality of distinct language lexicon models.

14. The method as defined in claim 13, further comprising generating confidence scores for each speech recognition output and determining a best model combination based upon the confidence scores.

15. A computer readable storage medium for generating speech recognition output using a plurality of distinct-language acoustic models, each based upon a different language, and at least one lexicon model, the storage medium comprising computer instructions for:

generating a first speech recognition output using a first model combination that combines one of the plurality of distinct-language acoustic models with the at least one lexicon model; and

in response to a threshold determination, generating a second speech recognition output using a second model combination that combines a different one of the plurality of distinct-language acoustic models with the at least one distinct-language lexicon model.

16. The computer readable storage medium as defined in claim 15, further comprising a computer instruction for determining a first confidence score for the first speech recognition output, and wherein the threshold determination is based upon a comparison of the first confidence score with a minimum acceptable score.

17. The computer readable storage medium as defined in claim 15, further comprising a computer instruction for generating confidence scores for each speech recognition output and determining a best model combination based upon the confidence scores.

18. The computer readable storage medium as defined in claim 15, wherein the at least one lexicon model comprises a plurality of distinct-language lexicon models, each based upon a different language.

19. The computer readable storage medium as defined in claim 18, further comprising a computer instruction for generating additional speech recognition outputs using different model combinations comprising different pairings of the plurality distinct-language acoustic models with the plurality of distinct language lexicon models.

20. The computer readable storage medium as defined in claim 19, further comprising computer instructions for generating confidence scores for each speech recognition output and determining a best model combination based upon the confidence scores.