System, method and computer program product for a distributed speech recognition tuning platform

• US 7,383,187 B2
• Filed: 01/24/2001
• Issued: 06/03/2008
• Est. Priority Date: 01/24/2001
• Status: Expired due to Term

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Bevocal Incorporated (Microsoft Corporation)
• Inventors: Damiba, Bertrand A
• Primary Examiner(s): MCFADDEN, SUSAN IRIS
• Application Number:
  US09/769,880
• Publication Number:
  US 20020138276A1
• Time in Patent Office: 2,687 Days
• Field of Search: 704/270, 704/260, 704/270.1, 704/275, 379/88.01, 707/6
• US Class Current: 704/270
• CPC Class Codes: G10L 15/06 Creation of reference templates; Training of speech recognition systems, e.g. adaptation to the characteristics of the speaker's voice G10L15/14 takes precedence
• CPC Class Codes: G10L 15/30 Distributed recognition, e.g. in client-server systems, for mobile phones or network applications

Abstract

A system, method and computer program product are provided for tuning a speech recognition process. Initially, a database of utterances is maintained. Thereafter, information associated with the utterances is collected utilizing a speech recognition process. Further, the utterances in the database are transmitted to a plurality of users utilizing a network. As such, transcriptions of the utterances in the database may be received from the users utilizing the network. In use, the speech recognition process may be tuned utilizing the information and the transcriptions.

Claims

1. A method for improving a speech recognition process, comprising:

maintaining a database of utterances;

collecting information associated with the utterances in the database utilizing a speech recognition process;

transmitting the utterances in the database to at least one user interface utilizing a network;

receiving transcriptions of the utterances in the database from the at least one user interface utilizing the network;

wherein a human is capable of utilizing the information and the transcriptions to improve a speech recognition application;

wherein the speech recognition process is improved by performing experiments based on the information;

wherein the information is selected from the group consisting of a dialog state, a gender of a speaker, and a date the utterances are wherein the at least one user interface includes a first icon for emitting a present utterance upon the selection thereof.

2. The method as recited in claim 1, wherein the network includes the Internet.

3. The method as recited in claim 2, wherein the transcriptions of the utterances are received from the at least one user interface using a network browser.

4. The method as recited in claim 1, wherein the information includes a recognition result.

5. The method as recited in claim 1, wherein the information is selected from the group consisting of a name of a grammar each utterance was recognized against, a name of an audio file on a disk, a directory path to the audio file, a size of the audio file, a session identifier, an index of each utterance, a recognition status, a recognition confidence associated with a recognition result, a recognition hypothesis, and an identification of a transcriber.

6. The method as recited in claim 1, wherein the information includes a name of a grammar each utterance was recognized against, a name of an audio file on a disk, a directory path to the audio file, a size of the audio file, a session identifier, an index of each utterance, the dialog state, a recognition status, a recognition confidence associated with a recognition result, a recognition hypothesis, the gender of a speaker, an identification of a transcriber, and the date the utterances are transcribed.

7. The method as recited in claim 1, wherein the utterances and the information are stored in the database, and the database is capable of being queried for results selected from the group consisting of a number of the utterances, a percentage of rejected utterances for a grammar, an average length of each utterance, a call volume in a predetermined range, a popularity of a grammar state, and a transcription management parameter.

8. The method as recited in claim 1, wherein the utterances and the information are stored in the database, and the database queried for results includes a number of the utterances, a percentage of rejected utterances for a grammar, an average length of each utterance, a call volume in a predetermined range, a popularity of a grammar state, and a transcription management parameter.

9. The method as recited in claim 1, wherein the speech recognition application is improved by performing experiments based on the information.

10. The method as recited in claim 1, wherein the at least one user interface includes additional icons for emitting previous and next utterances upon the selection thereof.

11. The method as recited in claim 1, wherein the at least one user interface includes a comment field for allowing a user to enter comments regarding a plurality of transcriptions.

12. The method as recited in claim 1, wherein the at least one user interface includes a hint menu for allowing a user to choose from a plurality of stings identified by the speech recognition process.

13. The method as recited in claim 12, wherein the hint menu allows the user to do a manual comparison between the utterances and results of the speech recognition process.

14. The method as recited in claim 1, wherein the information includes the dialog state.

15. The method as recited in claim 14, wherein the dialog state includes a context in a dialog flow.

16. The method as recited in claim 1, wherein the information includes the gender of the speaker.

17. The method as recited in claim 1, wherein the information includes the date the utterances are transcribed.

18. A computer program product embodied on a computer readable medium for improving a speech recognition process, comprising:

(a) computer code for maintaining a database of utterances;

(b) computer code for collecting information associated with the utterances in the database utilizing a speech recognition process;

(c) computer code for transmitting the utterances in the database to at least one user interface utilizing a network; and

(d) computer code for receiving transcriptions of the utterances in the database from the at least one user interface utilizing the network;

(e) wherein a human is capable of utilizing the information and the transcriptions to improve a speech recognition application;

wherein the speech recognition process is improved by performing experiments based on the information;

wherein the information is selected from the group consisting of a dialog state, a gender of a speaker, and a date the utterances are wherein the at least one user interface includes a string field for allowing a user to enter a string corresponding to each utterance.

19. A system including a tangible computer readable medium for improving a speech recognition process, comprising:

(a) logic for maintaining a database of utterances;

(b) logic for collecting information associated with the utterances in the database utilizing a speech recognition process,

(c) logic for transmitting the utterances in the database to at least one user interface utilizing a network;

(d) logic for receiving transcriptions of the utterances in the database from the at least one user interface utilizing the network;

(e) wherein a human is capable of utilizing the information and the transcriptions to improve a speech recognition application;

wherein the speech recognition process is improved by performing experiments based on the information;

wherein the information is selected from the group consisting of a dialog state, a gender of a speaker, and a date the utterances are wherein the at least one user interface includes a string field for allowing a user to enter a string corresponding to each utterance.