System and method for processing speech

• US 7,720,203 B2
• Filed: 06/01/2007
• Issued: 05/18/2010
• Est. Priority Date: 12/06/2004
• Status: Active Grant

• Current Assignee: Interactions, LLC
• Original Assignee: AT&T Intellectual Property I LP (AT&T, Inc.)
• Inventors: Bushey, Robert R., Knott, Benjamin A., Korth, Sarah, Martin, John M.
• Primary Examiner(s): Gauthier; Gerald
• Application Number:
  US11/809,817
• Publication Number:
  US 20070244697A1
• Time in Patent Office: 1,082 Days
• Field of Search: 379/88.01, 379/88.03, 379/88.04, 379/88.14, 701/202, 704/9, 704/231, 704/251, 704/257, 704/260, 704/275, 704/270, 704/1, 704/7, 704/252, 704/254, 704/270.1, 715/251, 715/530, 715/728, 715/762, 707/102, 709/217, 712/200
• US Class Current: 379/88.03
• CPC Class Codes: G10L 15/00 Speech recognition G10L17/00 takes precedence
• CPC Class Codes: H04M 2201/40 using speech recognition speech recognition per se G10L15/00
• CPC Class Codes: H04M 2203/2011 Service processing based on information specified by a party before or during a call, e.g. information, tone or routing selection
• CPC Class Codes: H04M 3/5166 in combination with interactive voice response systems or voice portals, e.g. as front-ends

Abstract

Systems and methods for processing speech are provided. A system may include an acoustic model to transform speech input into one or more word strings. The system may also include a semantic model to convert each of the one or more word strings into a detected object and a detected action. The system may also include a synonym table to determine a preferred object based on the detected object and to determine a preferred action based on the detected action.

Claims

1. A system to process speech, the system comprising:

an acoustic model to transform speech input into one or more word strings;

a semantic model to convert each of the one or more word strings into a detected object and a detected action; and

a synonym table storing data that can be processed to determine a preferred object based on the detected object and to determine a preferred action based on the detected action.

2. The system of claim 1, further comprising a processor to perform an act based on the preferred action and the preferred object.

3. The system of claim 1, further comprising a processor to route a caller to a desired destination based on the preferred action and the preferred object.

4. The system of claim 1, further comprising a processor to determine a dominant action based at least partially on the preferred action, and to determine a dominant object based at least partially on the preferred object.

5. The system of claim 1, wherein the preferred object and preferred action are determined based at least partially on geographic designation information related to the speech input.

6. The system of claim 1, wherein the acoustic model transforms the speech input into the one or more word strings based on detected phonemes.

7. The system of claim 1, wherein the acoustic model assigns probability values to the one or more word strings based on an evaluation of coherency of the one or more word strings.

8. The system of claim 1, wherein the semantic model parses the one or more word strings to identify nouns to determine the detected object and wherein the semantic model parses the one or more word strings to identify verbs to determine the detected action.

9. The system of claim 1, wherein the semantic model assigns a first probability value to the detected object and a second probability value to the detected action, wherein the first probability value and the second probability value indicate a likelihood that the detected object and the detected action reflect an intent of a speaker of the speech input.

10. A system to process speech, the system comprising:

a synonym table including a plurality of preferred objects associated with a plurality of detected objects and a plurality of preferred actions associated with a plurality of detected actions; and

a processor to receive a detected object based on speech input, to select a preferred object from the synonym table based on the detected object, to receive a detected action based on the speech input, and to select a preferred action from the synonym table based on the detected action.

11. The system of claim 10, wherein at least one of the plurality of detected objects of the synonym table includes a natural language phrase including at least one noun.

12. The system of claim 10, wherein at least one of the plurality of detected actions of the synonym table includes a natural language phrase including at least one verb.

13. The system of claim 10, wherein at least one of the plurality of detected objects of the synonym table includes a natural language phrase including at least one adjective and at least one noun.

14. The system of claim 10, wherein at least one of the plurality of detected actions of the synonym table includes a natural language phrase including at least one adverb and at least one verb.

15. The system of claim 10, wherein the processor determines a first confidence level associated with the preferred object based on the detected object, and wherein the processor determines a second confidence level associated with the preferred action based on the detected action.

16. The system of claim 10, wherein the processor determines a plurality of preferred objects based on the detected object, and selects a dominant object from the plurality of preferred objects.

17. The system of claim 10, wherein the processor determines a plurality of preferred actions based on the detected action, and selects a dominant action from the plurality of preferred actions.

18. The system of claim 10, further comprising an object-action routing module to route a call associated with the speech input, wherein the object-action routing module includes a plurality of call destinations related to object-action pairs.

19. The system of claim 18, wherein more than 80% of calls processed using the object-action routing module are successfully routed to a desired destination.