Application of user-specified transformations to automatic speech recognition results

• US 8,775,183 B2
• Filed: 06/12/2009
• Issued: 07/08/2014
• Est. Priority Date: 06/12/2009
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Hamaker, Jonathan E., Herold, Keith C.
• Primary Examiner(s): COLUCCI, MICHAEL C
• Application Number:
  US12/483,919
• Publication Number:
  US 20100318356A1
• Time in Patent Office: 1,852 Days
• Field of Search: 704/8, 704/9, 704/270.1, 704/270, 704/260, 704/251, 704/235, 704/232, 715/210, 715/205, 709/227, 709/206, 707/741, 705/80, 705/20, 341/51
• US Class Current: 704/251
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence

Abstract

Textual transcription of speech is generated and formatted according to user-specified transformation and behavior requirements for a speech recognition system having input grammars and transformations. An apparatus may include a speech recognition platform configured to receive a user-specified transformation requirement, recognize speech in speech data into recognized speech according to a set of recognition grammars; and apply transformations to the recognized speech according to the user-specified transformation requirement. The apparatus may further be configured to receive a user-specified behavior requirement and transform the recognized speech according to the behavior requirement. Other embodiments are described and claimed.

Claims

1. A computer-implemented method comprising:

receiving, from a client device, a user-specified behavior requirement in a speech recognition system wherein the user-specified behavior requirement changes a setting to an input speech recognition grammar specification prior to recognizing speech to select or de-select a built-in transformation from a set of built-in transformations of the speech recognition system to apply to recognized speech, and wherein the user-specified behavior requirement includes a semantic tag that specifies at least one of;

a proper name, a phone number, an address, an email address, an internet protocol address, and a web address;

encoding the received user-specified behavior requirement into an input speech recognition grammar specification for the speech recognition system;

receiving a request to recognize speech from an application;

recognizing the speech to form a recognized word sequence;

transforming the recognized word sequence according to the user-specified behavior requirement according to a semantic tag associated with the word sequence; and

providing the transformed sequence to the application.

2. The method of claim 1, comprising:

receiving a user-specified transformation requirement in the speech recognition system wherein the user-specified transformation requirement includes a semantic tag; and

transforming the recognized word sequence according to at least one of the user-specified behavior requirement or the user-specified transformation requirement according to a semantic tag associated with the word sequence.

3. The method of claim 2, wherein transforming the recognized word sequence comprises:

determining which transformations to apply to a sub-segment of the recognized word sequence according to a semantic tag associated with the sub-segment.

4. The method of claim 3, wherein the semantic tag further specifies at least one of:

a date, and a time.

5. The method of claim 1 wherein the user-specified behavior requirement comprises at least one of:

a grammar, an application program interface (API) call, an extensible markup language (XML) element, an XML schema, or speech recognition application parameters.

6. The method of claim 1, wherein the user-specified behavior requirement specifies a first transformation for a first portion of the input speech recognition grammar specification, and a second transformation for a second portion of the input speech recognition grammar specification.

7. A system for implementing formatted recognized speech, the system comprising:

a speech recognition platform configured to;

receive a user-specified behavior requirement, wherein the user-specified behavior requirement changes a setting to an input speech recognition grammar specification to select or de-select a built-in transformation from a set of built-in transformations of the speech recognition system;

receive a user-specified transformation requirement, wherein the user-specified transformation requirement specifies prior to recognizing speech to the speech recognition platform transformations to apply to recognized speech and wherein the user-specified transformation requirement includes a semantic tag that specifies at least one of;

a proper name, a phone number, an address, an email address, an internet protocol address, and a web address;

encode the user-specified transformation requirement into the input speech recognition grammar specification for the speech recognition platform;

recognize speech in speech data into recognized speech according to a set of recognition grammars including the input speech recognition grammar specification; and

apply transformations to the recognized speech according to the user-specified transformation requirement.

8. The system of claim 7, further comprising:

a client application configured to;

create a user-specified transformation requirement;

encode the created user-specified transformation requirement into the speech recognition grammar specification for the speech recognition platform; and

request speech recognition using the user-specified transformation requirement from the speech recognition platform.

9. The system of claim 7, wherein the user-specified behavior requirement includes a semantic tag that specifies at least one of:

a proper name, a phone number, an address, an email address, an internet protocol address, and a web address; and

the speech recognition platform further configured to;

encode the user-specified behavior requirement into the speech recognition grammar specification for the speech recognition platform; and

transform the recognized word sequence according to the user-specified behavior requirement.

10. The system of claim 7, wherein at least one of the user-specified transformation requirement or the user-specified behavior requirement comprises at least one of:

a grammar, an application program interface (API) call, an extensible markup language (XML) element, an XML schema, or speech recognition application parameters.

11. The system of claim 9, the speech recognition platform further configured to:

determine which transformations to apply to a sub-segment of the recognized speech according to a semantic tag associated with the sub-segment.

12. The system of claim 7, wherein the user-specified behavior requirement specifies a first transformation for a first portion of the input speech recognition grammar specification, and a second transformation for a second portion of the input speech recognition grammar specification.

13. An article comprising a storage medium containing instructions that if executed enable a system to:

receive a request to recognize speech from an application;

recognize the speech to form a recognized word sequence;

transform the recognized word sequence according to a user-specified behavior requirement encoded into an input speech recognition grammar specification, wherein the user-specified behavior requirement changes a setting prior to recognizing speech to the input speech recognition grammar to select or de-select a built-in transformation from a set of built-in transformations of a speech recognition system to apply to recognized speech, and wherein the user-specified behavior requirement includes a semantic tag that specifies at least one of;

a proper name, a phone number, an address, an email address, an internet protocol address, and a web address; and

provide the transformed sequence to the application.

14. The article of claim 13, further comprising instructions that if executed enable the system to transform the recognized word sequence according to at least one of the user-specified behavior requirement or a user-specified transformation requirement, wherein the user-specified transformation requirement includes a semantic tag that specifies at least one of:

a proper name, a phone number, an address, an email address, an internet protocol address, and a web address.

15. The article of claim 13, wherein at least one of the user-specified behavior requirement or a user-specified transformation requirement includes a semantic tag, and further comprising instructions that if executed enable the system to:

determine which transformations to apply to a sub-segment of the recognized word sequence according to a semantic tag associated with the sub-segment.

16. The article of claim 13, wherein the user-specified behavior requirement specifies a first transformation for a first portion of the input speech recognition grammar specification, and a second transformation for a second portion of the input speech recognition grammar specification.