×

Language recognition using a similarity measure

  • US 7,310,600 B1
  • Filed: 10/25/2000
  • Issued: 12/18/2007
  • Est. Priority Date: 10/28/1999
  • Status: Expired due to Fees
First Claim
Patent Images

1. A comparison apparatus comprising:

  • a receiver operable to receive an input signal;

    a recognition processor operable to compare said input signal with stored label models to generate a recognised sequence of labels in said input signal and confidence data representative of the confidence that the recognised sequence of labels is representative of said input signal; and

    a similarity measure calculator operable to compare said recognised sequence of labels received from said recognition processor with a stored sequence of annotation labels using a combination of i) predetermined confusion data which defines confusability between different labels, and ii) said confidence data received from the recognition processor and representative of the confidence that said received recognized sequence of labels is representative of the input signal, to provide a measure of the similarity between the recognised sequence of labels and the stored sequence of annotation labels,wherein said similarity measure calculator comprisesan aligner operable to align labels of the recognised sequence of labels with labels of the stored sequence of annotation labels to form a number of aligned pairs of labels;

    a comparator operable to compare the labels of each aligned pair of labels using said combination of said predetermined confusion data and said confidence data, to generate a comparison score representative of the similarity between the aligned pair of labels; and

    a combiner operable to combine the comparison scores for all the aligned pairs of labels to provide said similarity measure,wherein said comparator comprisesa first sub-comparator operable to compare, for each aligned pair, the recognised sequence label in the aligned pair with each of a plurality of labels taken from a set of predetermined labels using said confusion data and said confidence data to provide a corresponding plurality of intermediate comparison scores representative of the similarity between said recognised sequence label and the respective labels from the set;

    a second sub-comparator operable to compare, for each aligned pair, the stored sequence label in the aligned pair with each of said plurality of labels from the set using said confusion data and said confidence data to provide a further corresponding plurality of intermediate comparison scores representative of the similarity between said stored sequence label and the respective labels from the set; and

    a calculator operable to calculate said comparison score for the aligned pair by combining said pluralities of intermediate comparison scores.

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