Systems and methods for speech indexing

• US 8,326,631 B1
• Filed: 03/31/2009
• Issued: 12/04/2012
• Est. Priority Date: 04/02/2008
• Status: Active Grant

• Current Assignee: Verint Systems Incorporated
• Original Assignee: Verint Americas Incorporated (Verint Systems Incorporated)
• Inventors: Watson, Joseph Alva
• Primary Examiner(s): He, Jialong
• Application Number:
  US12/415,688
• Time in Patent Office: 1,344 Days
• Field of Search: 704/251, 704/270, 704/275
• US Class Current: 704/270
• CPC Class Codes: G06F 16/7844 using original textual content or text extracted from visual content or transcript of audio data
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 15/32 Multiple recognisers used in sequence or in parallel; Score combination systems therefor, e.g. voting systems
• CPC Class Codes: G10L 2015/025 Phonemes, fenemes or fenones being the recognition units

Abstract

A speech index for a recording or other representation of an audio signal containing speech is generated using a phonetic automatic voice recognition engine. A second speech index is also generated using a more accurate, but slower, automatic voice recognition engine such as a large vocabulary speech recognition (LVSR) engine. These two speech indexes are compared. The results of the comparison are then used to adjust certain parameters used by the phonetic engine while generating a speech index. The results may also be used to correct all or parts of the speech index generated by the phonetic automatic speech recognition engine.

Claims

1. A method of indexing speech, comprising:

associating a first phonetic sequence with a first position in an audio signal using a phonetic recognizer;

associating said first phonetic sequence to a first linguistic element based on a first parameter;

associating a second linguistic element with a second position in said audio signal using a large vocabulary speech recognizer (LVSR);

comparing said first position and said second position to determine a phrase window;

comparing said first linguistic element to said second linguistic element if said phrase window meets a first criteria; and

adjusting said first parameter based upon a result of said step of comparing said first linguistic element

wherein said step of associating said second linguistic element is performed on a lesser portion of said audio signal than said step of associating said first phonetic sequence with said first position;

wherein said step of associating said first phonetic sequence to said first linguistic element also associates said first linguistic element with a confidence value and said lesser portion of said audio signal is selected to correspond to said first linguistic element based upon said confidence value.

2. The method of claim 1, further comprising:

associating said first position with said second linguistic element.

3. The method of claim 1, further comprising:

associating said first position with said second linguistic element.

4. The method of claim 1 wherein said step of adjusting said first parameter comprises increasing a probability that said second linguistic element will be associated with said first phonetic sequence by said step of associating said first phonetic sequence to a first linguistic element based on said first parameter.

5. The method of claim 1, wherein said step of comparing further comprises:

correlating a second phonetic sequence associated with said second linguistic element with said first phonetic sequence.

6. A system for indexing speech, comprising:

a phonetic decoder that associates audio features of an audio signal with a first phonetic sequence at a first position in said audio signal;

a lexical interpreter that associates said first phonetic sequence with a first linguistic element based on a first parameter;

large vocabulary speech recognizer that associates a second linguistic element with a second position in said audio signal;

a speech index comparator that compares said first position and said second position to determine a phrase window;

and, said speech index comparator also compares said first linguistic element to said second linguistic element if said phrase window meets a first criteria; and

a parameter adjuster that adjusts said first parameter based upon a result of said speech index comparator

wherein said large vocabulary speech recognizer performs said association on a lesser portion of said audio signal than said phonetic decoder;

wherein said lexical interpreter also associates said first linguistic element with a confidence value and said lesser portion of said audio signal is selected to correspond to said first linguistic element based upon said confidence value.

7. The system of claim 6, further comprising:

an index updater that associates said first position with said second linguistic element.

8. The system of claim 6, further comprising:

an index updater that associates said first position with said second linguistic element.

9. The system of claim 6, wherein adjusting said parameter adjuster increases a probability that said second linguistic element will be associated with said first phonetic sequence by said lexical interpreter.

10. The system of claim 6, further comprising:

a phonetic sequence correlator that correlates a second phonetic sequence associated with said second linguistic element with said first phonetic sequence.

11. A program storage device readable by a machine, tangibly embodying a program of instructions executable by the machine to perform method steps for indexing speech, comprising:

associating a first phonetic sequence with a first position in an audio signal using a phonetic recognizer;

associating said first phonetic sequence to a first linguistic element based on a first parameter;

associating a second linguistic element with a second position in said audio signal using a large vocabulary speech recognizer a (LVSR);

comparing said first position and said second position to determine a phrase window;

comparing said first linguistic element to said second linguistic element if said phrase window meets a first criteria; and

,

adjusting said first parameter based upon a result of said step of comparing said first linguistic element

wherein said step of associating said second linguistic element is performed on a lesser portion of said audio signal than said step of associating said first phonetic sequence with said first position;

wherein said step of associating said first phonetic sequence to said first linguistic element also associates said first linguistic element with a confidence value and said lesser portion of said audio signal is selected to correspond to said first linguistic element based upon said confidence value.

12. The program storage device of claim 11, wherein the method further comprises:

associating said first position with said second linguistic element.

13. The program storage device of claim 11, wherein the method further comprises:

associating said first position with said second linguistic element.

14. The program storage device of claim 11 wherein said step of adjusting said first parameter comprises increasing a probability that said second linguistic element will be associated with said first phonetic sequence by said step of associating said first phonetic sequence to a first linguistic element based on said first parameter.