INTERNET BASED SPEECH RECOGNITION SYSTEM WITH DYNAMIC GRAMMARS

• US 20070185716A1
• Filed: 04/09/2007
• Published: 08/09/2007
• Est. Priority Date: 11/12/1999
• Status: Active Grant

• Current Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Original Assignee: Ian M. Bennett
• Inventors: Bennett, Ian
• Granted Patent: US 7,725,320 B2
• Time in Patent Office: Days
• US Class Current: 704/270.100
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

A speech-enabled WWW based computing system allows a user to interact with content associated with a web page and select items of interest using speech as a mode of input. Dynamic grammars can assist in the recognition operations to improve speed and comprehension.

Claims

1. A world-wide web (WWW) accessible natural language computing system comprising:

a speech recognition engine configured to distribute speech processing operations between a client device and a server device for processing an utterance for a speech based query;

said server device being configured to generate a recognized speech query using a grammar accessible to said server device;

a natural language routine executing on the server device and configured to process said recognized speech query to generate a natural language result in real-time;

an Internet accessible page coupled to the server device and having a list of items, wherein content for at least some of said list of items identified on said Internet accessible page is selectable by a user based on said natural language result;

a database coupled to the server device for storing answers which correspond to said content for said list of items on said Internet accessible page;

wherein a grammar used to recognize said speech based query can be varied between utterances and loaded dynamically as needed to recognize utterances associated with said content.

2. The WWW accessible natural language computing system of claim 1, wherein said Internet accessible page is adapted so that the user can navigate and locate information of interest using said speech query.

3. The WWW accessible natural language computing system of claim 1, wherein said list of items include products and/or services.

4. The WWW accessible natural language computing system of claim 1, wherein said list of items includes a list of frequently asked questions associated with a company'"'"'s products and/or services.

5. The WWW accessible natural language computing system of claim 1, wherein said system is further adapted to respond to a speech query concerning said list of items by returning a text or speech articulated response.

6. The WWW accessible natural language computing system of claim 1, further including an interactive character agent presented to the user for assisting in handling said speech query.

7. The WWW accessible natural language computing system of claim 1, wherein respective speech processing operations to be performed by the client platform and the server computing system are specified by an initialization routine.

8. The WWW accessible natural language computing system of claim 1 wherein the server device transfers speech related data using a hypertext transfer protocol (HTTP) and a format which includes an embedded NULL character.

9. The WWW accessible natural language computing system of claim 1, wherein the speech recognition engine stores calibration configuration data pertaining to calibrating speech and silence components of a speech utterance for a plurality of portable devices.

10. The WWW accessible natural language computing system of claim 1, wherein said speech based query is recognized by forming a concatenation of words and/or phrases derived from said speech query and using said concatenation as a search query for a database.

11. The WWW accessible natural language computing system of claim 1 wherein multiple speech grammars are available and selectable within the Internet accessible page, and such that speech input provided by the user for an item within the Internet accessible page using a first grammar dynamically controls which one of a plurality of second grammars is loaded for speech recognition of subsequent speech input by the user.

12. A World-Wide Web (WWW) accessible natural language computing system comprising:

a speech recognition engine configured to distribute speech processing operations between a client device and a server device for processing an utterance for a speech based query;

said server device being configured to generate a recognized speech query using a speech recognition grammar accessible to said server device;

a natural language routine executing on the server device and configured to process said recognized speech query to generate a natural language result in real-time;

wherein said natural language routine is configured to analyze phrases within said recognized speech query which are not in said speech recognition grammar;

an Internet accessible page coupled to the server device and having a list of items, wherein content for at least some of said list of items identified on said Internet accessible page is selectable by a user based on said natural language result;

a database coupled to the server device for storing answers which correspond to said content for said list of items on said Internet accessible page;

wherein a speech recognition grammar used to recognize said speech based query can be varied between utterances and loaded dynamically as needed to recognize utterances associated with said content.

13. The WWW accessible natural language computing system of claim 12 wherein said natural language routines analyzes a set of answers from said database to identify a best response.

14. The WWW accessible natural language computing system of claim 12 wherein said speech recognition grammar is based on words derived from a set of users within a geographical region served by said system.

15. A World-Wide Web (WWW) accessible natural language computing system comprising:

a speech recognition engine configured to distribute speech processing operations between a client device and a server device for processing an utterance for a speech based query;

said server device being configured to generate a recognized speech query using a speech recognition grammar accessible to said server device;

a natural language routine executing on the server device and configured to process said recognized speech query to generate a natural language result in real-time;

wherein said natural language routine is configured to analyze words and phrases in said recognized speech query which are not in said speech recognition grammar;

a WWW accessible page coupled to the server device and having a list of items accessible within a WWW browser, wherein content for at least some of said list of items identified on said WWW accessible page is selectable by a user based on said natural language result;

a database coupled to the server device for storing answers which correspond to said content for said list of items on said WWW accessible page;

wherein a speech recognition grammar used to recognize said speech based query can be varied between utterances and loaded dynamically as needed to recognize utterances associated with said content.