METHOD AND SYSTEM FOR PROCESSING MULTIPLE SPEECH RECOGNITION RESULTS FROM A SINGLE UTTERANCE

• US 20110161077A1
• Filed: 12/30/2010
• Published: 06/30/2011
• Est. Priority Date: 12/31/2009
• Status: Active Grant

• Current Assignee: Volt Delta Resources LLC (NewNet Communication Technologies LLC)
• Original Assignee: Volt Delta Resources LLC (NewNet Communication Technologies LLC)
• Inventors: Bielby, Gregory J.
• Granted Patent: US 9,117,453 B2
• Time in Patent Office: Days
• US Class Current: 704/231
• CPC Class Codes: G06F 40/58 Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 15/32 Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems
• CPC Class Codes: G10L 17/02 Preprocessing operations, e.g. segment selection; Pattern representation or modelling, e.g. based on linear discriminant analysis [LDA] or principal components; Feature selection or extraction
• CPC Class Codes: G10L 2015/228 of application context

Abstract

A method of and system for accurately determining a caller response by processing speech-recognition results and returning that result to a directed-dialog application for further interaction with the caller. Multiple speech-recognition engines are provided that process the caller response in parallel. Returned speech-recognition results comprising confidence-score values and word-score values from each of the speech-recognition engines may be modified based on context information provided by the directed-dialog application and grammars associated with each speech-recognition engine. An optional context database may be used to further reduce or add weight to confidence-score values and word-score values, remove phrases and/or words, and add phrases and/or words to the speech-recognition engine results. In situations where a predefined threshold-confidence-score value is not exceeded, a new dynamic grammar may be created. A set of n-best hypotheses of what the caller uttered is returned to the directed-dialog application.

Claims

1. A system comprising:

a directed-dialog-processor server having a directed-dialog-processor application executing thereon;

a speech-recognition-engine server having a plurality of parallel-operable speech-recognition-engine applications executing thereon;

a context database;

a multiple-recognition-processor server in data communication with the directed-dialog-processor server, the speech-recognition-engine server, and the context database and having a multiple-recognition-processor application executing thereon; and

wherein the multiple-recognition-processor server is operable, via the multiple-recognition-processor application, to;

receive context information and a forwarded caller response from the directed-dialog-processor application;

select a set of parallel-operable speech-recognition-engine applications from the plurality of parallel-operable speech-recognition-engine applications;

combine the context information with additional context information from the context database to form modified context information;

forward to each speech-recognition-engine application in the selected set the modified context information, the forwarded caller response, and a request to perform speech recognition of the forwarded caller response; and

receive from each speech-recognition-engine application in the selected set an n-best list comprising at least one confidence-score value and at least one word-score value.

2. The system of claim 1, wherein the selection of the set of speech-recognition-engine applications from the plurality of speech-recognition-engine applications comprises analyzing configuration files in data communication with the multiple-recognition processor.

3. The system of claim 1, wherein the at least one confidence-score value and at least one word-score value in each n-best list are modified by a weight-multiplier value based on the context information provided by the directed-dialog-processor application, thereby creating a modified n-best list.

4. The system of claim 3, wherein each modified n-best list is combined into a single, sorted combined n-best list.

5. The system of claim 4, wherein the at least one confidence-score value and at least one word-score value of the sorted combined n-best list are modified by comparing whether the phrases and words comprising the sorted combined n-best list are present in the context database.

6. The system of claim 5, wherein the sorted combined n-best list is re-sorted following modification of the at least one confidence-score value and the at least one word-score value.

7. The system of claim 6, wherein, responsive to a determination that a confidence-score value of the at least one confidence-score value in the sorted combined n-best list exceeds a predefined threshold confidence-score value, the multiple-recognition processor sets an acceptable status indicator to a value instructing the directed-dialog processor to accept the entry in the n-best list with the highest confidence-score value and forwards to the directed-dialog-processor the n-best list and the acceptance status indicator.

8. The system of claim 4, wherein, responsive to a determination that a confidence-score value of the at least one confidence-score value in the sorted combined n-best list does not exceed a predefined threshold confidence-score value, the multiple-recognition processor generates a dynamic grammar.

9. The system of claim 8, wherein the multiple-recognition processor forwards, to each selected speech-recognition-engine application, the dynamic grammar as the modified context information, the forwarded caller response, and a request to perform a speech-recognition of the forwarded caller response.

10. A method comprising:

(a) providing a processor;

(b) providing a memory interoperably coupled to the processor and having computer-readable processor instructions stored thereon;

(c) using the processor and the memory in combination with the computer-readable processor instructions to perform at least one of steps (d)-(h);

(d) receiving context information and a forwarded caller response from a directed-dialog-processor application executing on a directed-dialog-processor server;

(e) selecting a set of parallel-operable speech-recognition-engine applications from a plurality of parallel-operable speech-recognition-engine applications executing on a speech-recognition-engine server;

(f) combining the context information received in step (d) and additional context information present in a context database, thereby forming modified context information;

(g) forwarding modified context information, the forwarded caller response, and a request to perform speech recognition of the forwarded caller response to each speech-recognition-engine application selected in step (e);

(h) receiving from each speech-recognition-engine application of the set of speech recognition engine applications an n-best list comprising at least one confidence-score value and a at least one word-score value.

11. The method of claim 10, wherein step (e) is performed after analyzing configuration files in data communication with the multiple-recognition processor.

12. The method of claim 10, wherein, responsive to step (h), modifying the one or more confidence-score values and the at least one word-score value in each n-best list by a weight-multiplier value based on the context information provided by the directed-dialog-processor application, thereby creating a modified n-best list.

13. The method of claim 12, wherein each modified n-best list is combined into a single sorted combined n-best list.

14. The method of claim 13, wherein the at least one confidence-score value and at least one word-score value of the sorted combined n-best list are modified by comparing whether the phrases and words comprising the sorted combined n-best list are present in the context database.

15. The method of claim 14, wherein, responsive to the modification of the at least one confidence-score value and the at least one word-score value, re-sorting the sorted combined n-best list.

16. The method of claim 15, wherein, responsive to a determination that a confidence-score value of the at least one confidence-score value in the sorted combined n-best list exceeds a predefined threshold-confidence-score value, setting the value of an acceptable status indicator to a value instructing the directed-dialog processor to accept the entry in the n-best list with the highest confidence-score value and forwarding to the directed-dialog-processor the n-best list and the acceptance status indicator.

17. The method of claim 13, wherein, responsive to a determination that a confidence-score value of the at least one confidence-score value in the sorted combined n-best list does not exceed a predefined threshold-confidence-score value, generating a dynamic grammar.

18. The method of claim 17, wherein, responsive to the generation of a dynamic grammar, forwarding, to each selected speech-recognition-engine application, the dynamic grammar as the modified context information, the forwarded caller response, and a request to perform a speech-recognition of the forwarded caller response.

19. A computer-program product comprising a non-transitory computer-usable medium having computer-readable processor instructions embodied therein, the non-transitory computer-readable processor instructions adapted to be executed to implement a method comprising:

(a) providing a processor;

(b) providing a memory interoperably coupled to the processor and having computer-readable processor instructions stored thereon;

(c) using the processor and the memory in combination to perform at least one of steps (d)-(h);

(d) receiving context information and a forwarded caller response from a directed-dialog-processor application executing on a directed-dialog-processor server;

(e) selecting a set of parallel-operable speech-recognition-engine applications from a plurality of parallel-operable speech-recognition-engine applications executing on a speech-recognition-engine server;

(f) combining the context information received in step (d) and additional context information present in a context database, thereby forming modified context information;

(g) forwarding modified context information, the forwarded caller response, and a request to perform speech recognition of the forwarded caller response to each speech-recognition-engine application selected in step (e);

(h) receiving from each speech-recognition-engine application of the set of speech recognition-engine applications an n-best list comprising at least one confidence-score value and at least one word-score value.

20. The computer-program product of claim 19, wherein step (e) is performed after analyzing configuration files in data communication with the multiple-recognition processor.

21. The computer-program product of claim 19, wherein, responsive to step (h), modifying the at least one confidence-score value and the at least one word-score value in each n-best list by a weight-multiplier value based on the context information provided by the directed-dialog-processor application, thereby creating a modified n-best list.

22. The computer-program product of claim 21, wherein each modified n-best list is combined into a single sorted combined n-best list.

23. The computer-program product of claim 22, wherein the at least one confidence-score value and the at least one word-score value of the sorted combined n-best list are modified by comparing whether the phrases and words comprising the sorted combined n-best list are present in the context database.

24. The computer-program product of claim 23, wherein, responsive to the modification of the at least one confidence-score value and at least one word-score value, re-sorting the sorted combined n-best list.

25. The computer-program product of claim 24, wherein, responsive to a determination that a confidence-score value of the at least one confidence-score value in the sorted combined n-best list exceeds a predefined threshold-confidence-score value, setting the value of an acceptable status indicator to a value instructing the directed-dialog processor to accept the entry in the n-best list with the highest confidence-score value and forwarding to the directed-dialog-processor the n-best list and the acceptance status indicator.

26. The computer-program product of claim 22, wherein, responsive to a determination that a confidence-score value of the at least one confidence-score value in the sorted combined n-best list does not exceed a predefined threshold-confidence-score value, generating a dynamic grammar.

27. The computer-program product of claim 26, wherein, responsive to the generation of a dynamic grammar, forwarding, to each selected speech-recognition-engine application, the dynamic grammar as the modified context information, the forwarded caller response, and a request to perform a speech-recognition of the forwarded caller response.