TOOL AND FRAMEWORK FOR CREATING CONSISTENT NORMALIZATION MAPS AND GRAMMARS

• US 20100191519A1
• Filed: 01/28/2009
• Published: 07/29/2010
• Est. Priority Date: 01/28/2009
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Gedge, Nicholas J., Morton, Rachel I., Rahmel, Heiko W.
• Granted Patent: US 8,990,088 B2
• Time in Patent Office: Days
• US Class Current: 704/9
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/193 Formal grammars, e.g. finite state automata, context free grammars or word networks

Abstract

A runtime framework and authoring tool are provided for enabling linguistic experts to author text normalization maps and grammar libraries without requiring high level of technical or programming skills. Authors define or select terminals, map the terminals, and define rules for the mapping. The tool enables an author to validate their work, by executing the map in the same way the recognition engine does, causing consistency in results from authoring to user operations. The runtime is used by the speech engines and by the tools to provide consistent normalization for supported scenarios.

Claims

1. A method to be executed at least in part in a computing device for providing a speech processing technology authoring tool and framework, the method comprising:

enabling a user to specify at least two terminals through a user interface of the authoring tool executed by a computing device;

enabling the user to create a mapping between the at least two terminals;

enabling the user to specify a rule based on the mapping;

enabling to user to specify at least one test scenario for validating the rule;

performing the specified test to determine a performance of the rule in a speech processing environment;

presenting test results such that the user is enabled to modify the rule through the user interface; and

specifying an ordering of expected test results for validating a weighting of the rule.

2. The method of claim 1, further comprising:

enabling the user to select the rule to be exposed semantically;

enabling the user to select at least one function to generate semantics based on the rule;

adding the semantically exposed rule to a grammar; and

enabling testing of grammar rules and semantic processing.

3. The method of claim 2, wherein the at least one function is according to a common functions definition such that sharing of functions and rules across distinct languages and dialects is enabled.

4. The method of claim 1, further comprising:

sharing a runtime for applying the rule between speech processing engines to use a grammar that includes the rule such that speech processing consistency between applications employing the speech processing engines is achieved, wherein speech processing includes one of speech recognition and speech synthesis.

5. The method of claim 1, further comprising:

sharing a runtime with a development tool such that consistent speech recognition user experience is achieved by employing language models derived from creating the maps with the development tool and applying the maps to language model input training text.

6. The method of claim 1, the authoring tool abstracts underlying logic to a graphical user interface (GUI) for enabling specification of terminals, list rules, sequence rules, and grammar rules.

7. The method of claim 6, wherein the GUI employs at least one of textual input and graphics based interaction to enable the user to specify the terminals, the list rules, the sequence rules, and the grammar rules.

8. The method of claim 1, wherein the mapping is transducer based mapping.

9. The method of claim 1, wherein the rule is for at least one of:

text normalization, inverse text normalization, and grammars.

10. The method of claim 1, wherein the specified test is executable anytime against the mapping such that the user is enabled to validate the rule multiple times prior to addition to a grammar.

11. The method of claim 1, wherein the test scenario is specified based on at least one of:

a user selection from a default group of test scenarios and user definition of the test scenario, and wherein a predefined schema is employed for defining the test scenarios.

12. The method of claim 1, wherein the rule is presented by the user interface in one of:

a searchable and collapsible tree format and in a searchable ordered format.

13. A computing device for providing a speech processing technology authoring tool and framework, comprising:

a memory;

a processor coupled to the memory for executing the speech processing technology authoring tool configured to;

abstract an underlying logic to a GUI for enabling a user to specify at least one of;

terminals, list rules, sequence rules, and grammar rules, wherein the list rules and sequence rules are based on a user specified mapping between the terminals;

perform a plurality of test scenarios specified by one of a default list of test scenarios and the user to validate the list rules and the sequence rules;

present the user with a result of the test scenarios for enabling the user to modify the list rules and the sequence rules;

upon completion of the list rules and the sequence rules, enable expose a user selected portion of the list rules and sequence rules semantically employing common functions such that the completed rules are added to a grammar; and

perform a plurality of test scenarios specified by one of a default list of test scenarios and the user to validate the additions to the grammar.

14. The computing device of claim 13, wherein the authoring tool is further configured to share a runtime for applying the list rules and the sequence rules with speech processing engines through the framework such that speech processing consistency between applications employing the speech processing engines performing one of speech recognition and speech synthesis is achieved.

15. The computing device of claim 13, wherein authoring tool is further configured to perform normalization of a word based on an attribute of at least one other adjacent word to ensure an agreement of corresponding attributes of the words.

16. The computing device of claim 15, wherein the attributes are based on a predefined schema associated with the normalization.

17. The computing device of claim 13, wherein the authoring tool is further configured to provide a weighting of elements of a list rule.

18. A computer-readable storage medium with instructions stored thereon for specification of consistent normalization mapping and grammars, the instructions comprising:

enabling a user to specify terminals and rules based on a user specified mapping between the terminals;

performing a plurality of test scenarios specified by at least one of;

a user selection and modification from a default list of test scenarios;

providing results of the test scenarios and enabling the user to modify parameters of the terminals and the rules;

exposing completed rules semantically to a grammar using common functions such that the functions are to be shared among different languages and dialects; and

abstracting a structure of the rules into a map to enable reuse of the rules for similar languages by importing the rule structure for a new language model and localizing the terminals.

19. The computer-readable storage medium of claim 18, wherein the instructions further comprise:

providing the terminals, the rules, and the test scenarios employing at least one of;

a graphical scheme and a color scheme; and

enabling the user to select and modify the terminals, the rules, and the test scenarios through at least one of;

textual input and graphical interaction.

20. The computer-readable storage medium of claim 18, wherein the tested rules are exposed to a grammar for use in one of a build verification test process and a speech processing engine that shares a runtime with an authoring tool creating the rules for end-user experience consistency, wherein the speech processing engine is employed by applications including at least one from a set of word processing applications, presentation applications, browsing applications, navigational assistance applications, and communication applications, and wherein at least one of the applications are executed in a distributed manner over a plurality of computing devices communicating through a network.