Method for variable resolution and error control in spoken language understanding

• US 8,180,639 B2
• Filed: 05/06/2011
• Issued: 05/15/2012
• Est. Priority Date: 03/14/2007
• Status: Active Grant

• Current Assignee: Verint Americas Incorporated (Verint Systems Incorporated)
• Original Assignee: Speechcycle Incorporated (Synchronoss Technologies Incorporated)
• Inventors: Pieraccini, Roberto, Dayanidhi, Krishna
• Primary Examiner(s): Harper, Vincent P
• Application Number:
  US13/102,103
• Publication Number:
  US 20110208526A1
• Time in Patent Office: 375 Days
• Field of Search: 704/251, 704/257
• US Class Current: 704/251
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

A method for variable resolution and error control in spoken language understanding (SLU) allows arranging the categories of the SLU into a hierarchy of different levels of specificity. The pre-determined hierarchy is used to identify different types of errors such as high-cost errors and low-cost errors and trade, if necessary, high cost errors for low cost errors.

Claims

1. A method of minimizing the effect of errors in an automated spoken language recognition system comprising the steps of:

receiving voice utterances from a speaker;

recognizing a received voice utterance and classifying the recognized voice utterance in one of predetermined hierarchical spoken language categories;

assigning a confidence level to the classification of said received voice utterance;

determining if the confidence level of the classification is below a predetermined threshold; and

selecting a category of lesser specificity and superior to said classified category in the hierarchy if said confidence level is below said predetermined threshold.

2. The method of claim 1, wherein a confidence level below said predetermined threshold is indicative of an erroneous classification of the utterance.