System, method and computer program product for genre-based grammars and acoustic models in a speech recognition framework

• US 20020169604A1
• Filed: 03/09/2001
• Published: 11/14/2002
• Est. Priority Date: 03/09/2001
• Status: Abandoned Application

• Current Assignee: Bevocal Incorporated (Microsoft Corporation)
• Original Assignee: Bevocal Incorporated (Microsoft Corporation)
• Inventors: Podesva, Robert J., Damiba, Bertrand A., Guerra, Lisa M.
• Application Number:
  US09/802,663
• Publication Number:
  US 20020169604A1
• Time in Patent Office: Days
• US Class Current: 704/231
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 15/19 Grammatical context, e.g. disambiguation of the recognition hypotheses based on word sequence rules

Abstract

A system, method and computer program product are provided for genre-based speech recognition. Initially, utterances are received from a user. Thereafter, a genre associated with the user is determined based on information independent from the utterances of the user. At least one acoustic model and/or grammar may then be selected based on the genre determination. Accordingly, the utterances may be recognized utilizing the selected acoustic model(s) and/or grammar(s) for the purpose of providing a service to the user.

Claims

1. A method for genre-based speech recognition, comprising the steps of:

(a) receiving utterances from a user;

(b) determining a genre associated with the user based on information independent from the utterances of the user;

(c) selecting acoustic models based on the genre determination; and

(d) recognizing the utterances utilizing the selected acoustic models for the purpose of providing a service to the user.

2. The method as recited in claim 1, wherein the genre includes gender.

3. The method as recited in claim 1, wherein the genre includes a location of the user.

4. The method as recited in claim 1, wherein the genre includes a medium by which the user is communicating the utterances.

5. The method as recited in claim 4, wherein the medium includes a wireless medium.

6. The method as recited in claim 1, wherein the acoustic models involve speech pitch and intensity of the utterances received from the user.

7. The method as recited in claim 1, wherein the genre is determined based on information collected from the user.

8. The method as recited in claim 7, wherein the information is extracted from a call description record.

9. The method as recited in claim 7, wherein the information is extracted prior to the utterances being received from the user.

10. The method as recited in claim 7, wherein the information is inferred from a history associated with the user.

11. The method as recited in claim 7, wherein the information is entered manually by the user.

12. The method as recited in claim 11, wherein the information is entered utilizing a computer coupled to a network.

13. The method as recited in claim 1, wherein the information is entered during a speech tuning process prior to the receipt of the utterances.

14. The method as recited in claim 13, wherein the information is entered manually by a transcriber.

15. A computer program product for genre-based speech recognition, comprising:

(a) computer code for receiving utterances from a user;

(b) computer code for determining a genre associated with the user based on information independent from the utterances of the user;

(c) computer code for selecting acoustic models based on the genre determination; and

(d) computer code for recognizing the utterances utilizing the selected acoustic models for the purpose of providing a service to the user.

16. A method for genre-based speech recognition, comprising the steps of:

(a) receiving utterances from a user;

(b) determining a genre associated with the user based on information independent from the utterances of the user;

(c) selecting grammars based on the genre determination; and

(d) recognizing the utterances utilizing the selected grammars for the purpose of providing a service to the user.