Application services interface to ASR

• US 9,196,250 B2
• Filed: 11/16/2012
• Issued: 11/24/2015
• Est. Priority Date: 11/16/2012
• Status: Active Grant

• Current Assignee: Blackberry Limited
• Original Assignee: 2236008 Ontario Inc. (Blackberry Limited)
• Inventors: Fry, Darrin Kenneth John
• Primary Examiner(s): VO, HUYEN X
• Application Number:
  US13/679,654
• Publication Number:
  US 20140142945A1
• Time in Patent Office: 1,103 Days
• Field of Search: 704/231, 704/235, 704/251, 704/270, 704/275, 704/270.1, 704/255, 704/257, 704238-240, 704/246, 704/247, 704250-252, 704/254
• US Class Current: 1/1
• 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

Abstract

An application services interface system includes an automatic speech recognition control application program interface that receives a request for a recognition session from an application-based automatic speech recognition controller. An automatic speech recognition control engine directs the performance of an automatic speech recognition module. The automatic speech recognition module compares a spoken utterance to a vocabulary of active grammars to generate recognition results through limited data interchanges or exchanges.

Claims

1. An application services interface system comprising:

a processor;

an automatic speech recognition control application program interface that receives a request for a recognition session from an application-based automatic speech recognition controller;

an automatic speech recognition control engine in communication with the automatic speech recognition control application program interface directs the performance of an automatic speech recognition module; and

the automatic speech recognition module comprises two or more synchronous processing modules delivering a combination of grammar-based ASR recognition results and natural language-based ASR recognition results and compares a spoken utterance to a vocabulary of active grammars to generate recognition results;

where the recognition results are generated by the two or more synchronous processing modules in an ASR session with limited callbacks between the automatic speech recognition control engine and the application-based automatic speech recognition controller.

2. The application services interface system of claim 1 where the recognition results are generated in a single recognition session with no callbacks between the automatic speech recognition control engine and the application-based automatic speech recognition controller by using the two or more synchronous processing modules to deliver the combination of grammar-based ASR recognition results and natural language-based ASR recognition results in the single session.

3. The application services interface system of claim 1 where the recognition results are generated through a limited data interchange between the automatic speech recognition control engine and the application-based automatic speech recognition controller by using the two or more synchronous processing modules.

4. The application services interface system of claim 1 where the recognition results comprise agnostic data.

5. The application services interface system of claim 1 where the automatic speech recognition module comprises a natural language-based speech engine.

6. The application services interface system of claim 5 where the automatic speech recognition module further comprises a grammar-based speech engine.

7. The application services interface system of claim 1 where the automatic speech recognition control engine maintains state by monitoring the details of the state of a speech recognition.

8. The application services interface system of claim 7 where the application-based automatic speech recognition controller is stateless.

9. The application services interface system of claim 1 where the automatic speech recognition module or the automatic speech recognition control engine controls the duration of the ASR session.

10. The application services interface system of claim 1 where the ASR session is subject to a predetermined time constraint.

11. The application services interface system of claim 1 where a speech application program interface includes the application-based automatic speech recognition controller and further includes ASR logic in which a user interacts, and where the automatic speech recognition module includes the ASR logic that recognizes speech.

12. The application services interface system of claim 1 where the automatic speech recognition module resides on a cloud based computing platform.

13. The application services interface system of claim 1 where the automatic speech recognition module recognizes speech by comparing a spoken utterance to a natural language-based vocabulary and a grammar-based vocabulary.

14. The application services interface system of claim 1 where the automatic speech recognition control engine and the automatic speech recognition module each comprise a computing thread executed by a multitasking processor.

15. An application services interface system comprising:

a processor;

an automatic speech recognition control application program interface that receives a request for a recognition session from an application-based automatic speech recognition controller;

an automatic speech recognition control engine in communication with the automatic speech recognition control application program interface directs the performance of an automatic speech recognition module; and

the automatic speech recognition module comprises two or more synchronous processing modules compares a spoken utterance to a vocabulary of active grammars to generate recognition results;

where the recognition results are generated by the two or more synchronous processing modules in an ASR session;

wherein each of the two or more synchronous processing modules of the automatic speech recognition module comprises a plurality of speech engines that are run simultaneously in which content from a request for a recognition result is divided between the plurality of speech engines of each of the synchronous processing modules.

16. An application services interface system comprising:

a processor;

an automatic speech recognition control application program interface that receives a request for a recognition session from an application-based automatic speech recognition controller;

an automatic speech recognition control engine in communication with the automatic speech recognition control application program interface directs the performance of an automatic speech recognition module; and

the automatic speech recognition module comprises two or more synchronous processing modules compares a spoken utterance to a vocabulary of active grammars to generate recognition results;

wherein the two or more synchronous processing modules deliver a combination of grammar-based ASR recognition results and natural language-based ASR recognition results in a single ASR session and without callbacks.

17. An application services interface comprising:

a processor;

an automatic speech recognition control application program interface that receives a request for a recognition session from an application-based automatic speech recognition controller and transmits multi-processed recognition results to an application or a designated address; and

an automatic speech recognition control engine comprising two or more synchronous processing modules in communication with the automatic speech recognition control application program interface that directs the performance of a plurality of speech engines; and

the plurality of speech engines comprise at least one natural language-based speech engine and at least one grammar-based speech engine that compare a spoken utterance to a first vocabulary and then compares a first recognition result to active grammars that comprise a plurality of selected words that are each mapped to one of a plurality of domain specific words to generate the multi-processed recognition results;

where the multi-processed recognition results are generated in synchronism in an ASR session.

18. The application services interface of claim 17 where the multi-processed recognition results are generated using the least one natural language-based speech engine and the at least one grammar-based speech engine in succession.

19. A speech recognition process comprising:

receiving a request for a recognition session from an application-based automatic speech recognition controller that comprises two or more synchronous processing modules that allow for limited callbacks;

comparing a spoken utterance to a first vocabulary to generate a first speech recognition result;

comparing the first speech recognition result to a plurality of selected words that are each mapped to one of a plurality of domain specific words to generate multi-processed recognition results comprising a combination of grammar-based ASR recognition results and natural language-based ASR recognition results; and

transmitting the multi-processed recognition results to the source of the request for the recognition session or to an address;

where the multi-processed recognition results are generated in synchronism in a single ASR session.

20. The speech recognition process of claim 19 where the recognition results are generated by the two or more synchronous processing modules through a single data interchange between an automatic speech recognition control engine and a remote application-based automatic speech recognition controller.

21. The speech recognition process of claim 19 where the where the multi-processed recognition results comprise agnostic data.

22. An application services interface system comprising:

a processor;

an automatic speech recognition control application program interface that receives a request for a recognition session from an application-based automatic speech recognition controller;

an automatic speech recognition control engine in communication with the automatic speech recognition control application program interface directs the performance of an automatic speech recognition module; and

the automatic speech recognition module comprising at least one natural language-based speech engine and at least one grammar-based speech engine generating recognition results in synchronism that compares a spoken utterance to a vocabulary of active grammars to generate recognition results;

where the recognition results are generated in a single ASR session between the automatic speech recognition control engine and the application-based automatic speech recognition controller; and

where the application-based automatic speech recognition controller is automatically configured to the automatic speech recognition control engine when the application-based automatic speech recognition controller connects to the automatic speech recognition control application program interface.

23. The application services interface system of claim 22 where the automatic speech recognition control engine is further configured to recognize a speaker'"'"'s Intents.

24. The application services interface system of claim 22 where the automatic speech recognition control engine is further configured to receive Intents from the application-based automatic speech recognition controller.

25. The application services interface system of claim 22 where the automatic speech recognition control engine is pre-programmed with Intents.

26. The application services interface system of claim 22 where the application-based automatic speech recognition controller and the automatic speech recognition control engine are compliant with a Plug and Play specification.