Bilateral speech system

• US 6,430,531 B1
• Filed: 02/01/2000
• Issued: 08/06/2002
• Est. Priority Date: 02/04/1999
• Status: Expired due to Term

• Current Assignee: Cufer Asset Ltd LLC (Intellectual Ventures LLC)
• Original Assignee: Soliloquy, Inc.
• Inventors: Polish, Nathaniel
• Primary Examiner(s): Dorvil, Richemond
• Application Number:
  US09/495,722
• Time in Patent Office: 917 Days
• Field of Search: 704/275, 704/270, 704/270.1, 704/257, 704/254, 704/255, 704/249, 704/200
• US Class Current: 704/257
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

A speech processor that both speaks and listens uses speech recognizers as well as speech synthesizers to allow a user to engage in a verbal dialog with a data base. An element of working memory holds whatever is the context of the dialog so that the system can respond to successive requests or statements with greater and greater specificity.

Claims

1. A speech recognition system comprising:

an input device configured to receive a first statement in a natural language;

a system state controller configured to provide grammatical data to the input device, wherein the input device is further configured to generate a first query based on the first statement and the grammatical data;

a database configured to generate first information based on the first query;

an output device configured to generate a first answer in the natural language based on the first information; and

the input device further configured to receive a second statement in the natural language, the second statement comprising a pronoun, and configured to resolve the antecedent for the pronoun based on a context provided at least in part by the first statement.

3. The system of claim 1 or 2, wherein the system state controller comprises:

a memory bank configured to store the first query, the first information and the first answer.

4. The system of claim 3 wherein the input device comprises:

a speech recognizer configured to receive the first statement and configured to convert the first statement into a plurality of phonemes.

5. The system of claim 4 wherein the input device further comprises:

a first language model configured to generate a plurality of parsing tokens based on the plurality of phonemes and the grammatical data.

6. The system of claim 5 wherein the input device further comprises:

a query generator configured to generate the first query based on the plurality of parsing tokens.

7. The system of claim 1 or 2, wherein the first query is in Structured Query Language.

8. The system of claim 1 or 2, wherein the database is further configured to generate second information based on the second query;

and

the output device is further configured to generate a second answer in the natural language based on the second information.

9. The system of claim 8 further comprising:

a device controller configured to carry out a command from the system state controller, wherein the system state controller is further configured to generate the command based on at least one of the first information, the second information, the first answer and the second answer.

10. The system of claim 9 wherein the device controller is further configured to generate output to a user that includes at least one of the first answer and the second answer.

11. The system of claim 8 wherein the second information is a subset of the first information.

12. The system of claim 8 wherein the output device comprises:

a synthesizer configured to convert the second answer to a voice message.

2. A speech recognition system comprising:

an input device configured to receive a first statement in a natural language;

a system state controller configured to provide grammatical data to the input device, wherein the input device is further configured to generate a first query based on the first statement and the grammatical data;

a database configured to generate first information based on the first query;

an output device configured to generate a first answer in the natural language based on the first information; and

the input device further configured to receive a second statement in the natural language, the second statement comprising a homonym, and configured to disambiguate the homonym based on a context provided at least in part by the first statement.