System and method for providing a compensated speech recognition model for speech recognition

• US 7,451,085 B2
• Filed: 10/01/2001
• Issued: 11/11/2008
• Est. Priority Date: 10/13/2000
• Status: Expired due to Fees

• Current Assignee: Interactions, LLC
• Original Assignee: AT&T Intellectual Property II LP (AT&T, Inc.)
• Inventors: Pathasarathy, Sarangarajan, Rose, Richard C., Narayanan, Shrikanth Sambasivan, Rosenberg, Aaron Edward
• Primary Examiner(s): Wozniak, James S
• Application Number:
  US09/966,259
• Publication Number:
  US 20020059068A1
• Time in Patent Office: 2,598 Days
• Field of Search: 704/251, 704/275, 704243-244, 704/257, 704/256, 704/224, 704/270, 704/246, 704/233, 704/238, 704/222, 704/236, 704/270.1, 340/905
• US Class Current: 704/244
• CPC Class Codes: G10L 15/04 Segmentation; Word boundary detection
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 15/197 Probabilistic grammars, e.g. word n-grams
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/0631 Creating reference templates; Clustering

Abstract

An automatic speech recognition (ASR) system and method is provided for controlling the recognition of speech utterances generated by an end user operating a communications device. The ASR system and method can be used with a communications device that is used in a communications network. The ASR system can be used for ASR of speech utterances input into a mobile device, to perform compensating techniques using at least one characteristic and for updating an ASR speech recognizer associated with the ASR system by determined and using a background noise value and a distortion value that is based on the features of the mobile device. The ASR system can be used to augment a limited data input capability of a mobile device, for example, caused by limited input devices physically located on the mobile device.

Claims

1. An automatic speech recognition system, comprising:

a memory that stores a user profile having data related to user vocal information and value associated with a probability of the user being in a particular acoustic environment based on a time of day;

a controller coupled with the memory that receives the user profile and then compensates at least one speech recognition model based on the user profile;

a communication device that receives speech utterances from the user over a network; and

a speech recognizer that recognizes the speech utterances by using the at least one compensated speech recognition model.

2. The automatic speech recognition system according to claim 1, wherein user profile further includes transducer data related to a distortion value related to a transducer of a mobile communications device.

3. The automatic speech recognition system according to claim 1, wherein the particular acoustic environmental data includes a background noise value that corresponds to an operating environment of a mobile communications device.

4. The automatic speech recognition system according to claim 1, wherein the vocal information includes a distortion value related to the user associated with a mobile communications, device.

5. The automatic speech recognition system according to claim 1, wherein a personal computer is used provide the data of the particular acoustic environmental.

6. The automatic speech recognition system according to claim 1, wherein a personal digital assistant is used to provide the data of the particular acoustic environmental.

7. The automatic speech recognition system according to claim 1, wherein the data of the particular acoustic environmental is provided through a satellite communications system.

8. The automatic speech recognition system according to claim 1, wherein the speech recognizer is a network server using a hidden Markov model.

9. The automatic speech recognition system according to claim 1, wherein the controller is a network server that includes a pronunciation circuit, an environment-transducer-speaker circuit and a feature space circuit.

10. The automatic speech recognition system according to claim 8, wherein the network server updates the at least one speech recognition model and a pronunciation model to reflect a specific type of communications device.

11. The automatic speech recognition system according to claim 1, wherein the memory further stores personal account information that includes administrative information relating to the user.

12. The automatic speech recognition system according to claim 1, wherein the communications device can be configured by the user to select a specific speech recognition network.

13. A controller used in an automatic speech recognition system, comprising:

a receiving section that receives speech utterances over a network from a user;

a first section that determines user profile data related to user vocal information and value associated with a probability of the user being in a particular acoustic environment based on a time of day; and

a second section that compensates a speech recognition model for recognizing the speech utterances based on the user profile data.

14. The controller according to claim 13, wherein the controller identifies a user by a radio frequency identification tag.

15. The controller according to claim 13, wherein the acoustic environmental data is determined using at least one microphone in the user'"'"'s environment.

16. The controller according to claim 13, wherein the acoustic environmental data is determined using a plurality of microphones that are selectively initiated as the user walks in between the plurality of microphones.

17. The controller according to claim 13, where the user profile data further includes transducer data related to a distortion value based on a difference between an actual transducer in the mobile device and a response characteristic of a transducer used to train the speech recognition model.

18. The controller according to claim 13, wherein the vocal information represents a variability that exists in vocal tract shapes among speakers of a group.

19. The controller according to claim 13, wherein the controller communicates with a memory that stores various acoustic environmental models and various features of a specific type of mobile device.

20. The controller according to claim 19, wherein a third section stores personal account information for each end user.

21. A method of using an automatic-speech recognition system, comprising:

speech utterances over a network;

determining user profile data related to user vocal information and a value associated with a probability of the user being in a particular acoustic environment based on a time of day;

compensating a speech recognition model based on the user profile data; and

recognizing the speech utterances using the compensated speech recognition model.

22. The method according to claim 21, wherein the user profile further includes transducer data related to a distortion value related to a transducer used in a mobile device.

23. The method according to claim 22, wherein the user profile further includes data related to the acoustic environmental data includes a background noise value that corresponds to an operating environment of a mobile communications device.

24. The method according to claim 21, wherein the data of the particular acoustic environmental is received from a cellular telephone.

25. The method according to claim 21, wherein the data of the particular acoustic environmental is received from a personal digital assistant.

26. The method according to claim 21, wherein the data of the particular acoustic environmental is received via a satellite communications system.

27. The method according to claim 21, wherein the speech recognition model is a hidden Markov model.

28. The method according to claim 23, wherein determining the acoustic environmental data is performed using a network server.

29. The method according to claim 23, wherein the acoustic environmental data is determined using at least one microphone in the user'"'"'s environment.

30. The method according to claim 22, wherein the user profile includes data related to a transducer and a distortion value is determined based on a difference between an actual transducer in the mobile device and a response characteristic of a transducer used to train the speech recognition model.

31. The method according to claim 21, further comprising updating the speech recognition model and a pronunciation model to reflect a specific type of mobile communications device.

32. The method according to claim 21, further comprising configuring the communications device to select a specific speech recognition network.