×

Automating tuning of speech recognition systems

  • US 7,203,644 B2
  • Filed: 12/31/2001
  • Issued: 04/10/2007
  • Est. Priority Date: 12/31/2001
  • Status: Expired due to Fees
First Claim
Patent Images

1. A speech recognition system, comprising:

  • at least one recognizer to produce output signals from audio input signals based at least in part on speech models and a grammar file, the grammar file including at least one command syntax;

    a feedback module to generate feedback data, the feedback module modifying the speech models and the grammar file based on the feedback data to improve the performance of the at least one recognizer; and

    a controller adaptable to select one recognizer based at least in part on the feedback data from the at least one recognizer for an input utterance, the selected recognizer performing most accurately for the input utterance among the at least one recognizer.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×