System and method for processing speech recognition

• US 7,904,294 B1
• Filed: 04/09/2007
• Issued: 03/08/2011
• 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: Rose, Richard C., Pathasarathy, Sarangarajan, Rosenberg, Aaron Edward, Narayanan, Shrikanth Sambasivan
• Primary Examiner(s): ALBERTALLI, BRIAN LOUIS
• Application Number:
  US11/697,935
• Time in Patent Office: 1,429 Days
• Field of Search: 704/238, 704/256, 704/257, 704/231, 704/255, 704/2, 704/251, 704/220, 704/246, 704/240, 704/270, 704/275
• US Class Current: 704/238
• 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 mobile 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. A system for rescoring the results of automatic speech recognition, comprising:

a processor;

a first module configured to control the processor to generate a plurality of lattices for received speech utterances associated with filling in a plurality of user selectable data fields in a user interface;

a second module configured to control the processor to concatenate the plurality of lattices into a single concatenated lattice; and

a third module configured to control the processor to apply at least one language model to the single concatenated lattice for determining a relationship between lattices of the plurality of lattices.

2. The system for rescoring the results of automatic speech recognition according to claim 1, further comprising a fourth module configured to control the processor to generate a confidence score, after applying the at least one speech recognition model, and to control the processor to determine whether the plurality of lattices are acceptable.

3. The system of rescoring the results of automatic speech recognition according to claim 2, wherein the confidence score is compared to a predetermined value in order to determine whether to repeat automatic speech recognition.

4. The system of rescoring the results of automatic speech recognition according to claim 3, further comprising a fifth module configured to control the processor to repeat the automatic speech recognition process if the confidence score is less than the predetermined value.

5. The system of rescoring the results of automatic speech recognition according to claim 1, wherein the rescoring is performed after a speech recognition model has been compensated to reflect acoustic environmental data and transducer data.

6. The system of rescoring the results of automatic speech recognition according to claim 1, wherein the rescoring the automatic speech recognition is used in at least one of a mobile communications system and a satellite communications system.

7. A rescoring method used in an automatic speech recognition system, comprising:

generating a plurality of lattices using a speech recognition model associated with the automatic speech recognition system, each lattice corresponding to a received speech utterance associated with a user providing input to one of a plurality of user selectable data fields in a user interface; and

concatenating the plurality of lattices into a single concatenated lattice and applying said speech recognition model to the single concatenated lattice.

8. The rescoring method of claim 7, wherein the speech utterances are received from one of a mobile device, a personal digital assistant and a satellite communications system.

9. The rescoring method of claim 7, wherein the speech utterances are received from a personal digital assistant.

10. The rescoring method of claim 7, wherein the speech recognition model is a hidden Markov model.

11. The rescoring method of claim 7, wherein the speech recognition system is a network server.

12. A non-transitory computer-readable storage medium storing a computer program for controlling a computing device to perform automatic speech recognition, the instructions comprising:

generating a plurality of lattices for receiving speech utterances associated with filling in plurality of user selectable data fields in a user interface;

concatenating a plurality of lattices into a single concatenated lattice; and

applying at least one language model to the single concatenated lattice for determining a relationship between lattices of the plurality of lattices.

13. The non-transitory computer-readable storage medium of claim 12, wherein the instructions further comprise generating a confidence score, after applying the at least one speech recognition module, to determine whether the plurality of lattices are acceptable.

14. The non-transitory computer-readable storage medium of claim 13, wherein the confidence score is compared to a predetermined value in order to determine whether to repeat automatic speech recognition.

15. The non-transitory computer-readable storage medium of claim 14, wherein the instructions further comprise repeating the automatic speech recognition process if the confidence score is less than a predetermined value.

16. The non-transitory computer-readable storage medium of claim 12, wherein the rescoring is performed after a speech recognition model has been compensated to reflect acoustic environmental data and transducer data.

17. The non-transitory computer-readable storage medium of claim 12, wherein rescoring the automatic speech recognition is used in at least one of a mobile communications system and a satellite communication system.