Method and System for Real-Time Keyword Spotting for Speech Analytics

• US 20140025379A1
• Filed: 07/20/2012
• Published: 01/23/2014
• Est. Priority Date: 07/20/2012
• Status: Active Grant

• Current Assignee: Genesys Telecommunications Laboratories Incorporated (Genesys Cloud Services Incorporated)
• Original Assignee: Interactive Intelligence Incorporated (Genesys Cloud Services Incorporated)
• Inventors: Ganapathiraju, Aravind, Iyer, Ananth Nagaraja
• Granted Patent: US 9,672,815 B2
• Time in Patent Office: Days
• US Class Current: 704/255
• CPC Class Codes: G10L 15/08 Speech classification or search
• CPC Class Codes: G10L 15/142 Hidden Markov Models [HMMs]
• CPC Class Codes: G10L 2015/088 Word spotting

Abstract

A system and method are presented for real-time speech analytics in the speech analytics field. Real time audio is fed along with a keyword model, into a recognition engine. The recognition engine computes the probability of the audio stream data matching keywords in the keyword model. The probability is compared to a threshold where the system determines if the probability is indicative of whether or not the keyword has been spotted. Empirical metrics are computed and any false alarms are identified and rejected. The keyword may be reported as found when it is deemed not to be a false alarm and passes the threshold for detection.

Claims

1. A computer-implemented method for spotting predetermined keywords in an audio stream, comprising the steps of:

a) developing a keyword model for the predetermined keywords;

b) comparing the keyword model and the audio stream to spot probable ones of the predetermined keywords;

c) computing a probability that a portion of the audio stream matches one of the predetermined keywords from the keyword model;

d) comparing the computed probability to a predetermined threshold;

e) declaring a potential spotted word if the computed probability is greater than the predetermined threshold;

f) computing further data to aid in determination of mismatches;

g) using the further data to determine if the potential spotted word is a false alarm; and

h) reporting spotted keyword if a false alarm is not identified at step (g).

2. The method of claim 1, wherein step (a) comprises concatenating phoneme hidden Markov models of predetermined keywords.

3. The method of claim 1, wherein step (a) comprises:

a.1) creating a pronunciation dictionary that defines a sequence of phonemes for each of the predetermined keywords;

a.2) creating an acoustic model that statistically models a relation between textual properties of the phonemes for each of the predetermined keywords and spoken properties of the phonemes for each of the predetermined keywords; and

a.3) concatenating acoustic models for the sequence of phonemes for each of the predetermined keywords.

4. The method of claim 3, wherein step (a.2) comprises creating a set of Gaussian mixture models.

5. The method of claim 3, wherein step (a.2) comprises creating the acoustic model selected from the group consisting of:

context-independent model, context-dependent model, and triphone model.

6. The method of claim 1, wherein step (b) comprises:

b.1) converting the audio stream into a sequence of spectral features; and

b.2) comparing the keyword models to the sequence of spectral features.

7. The method of claim 6, wherein step (b.1) comprises:

b.1.1) converting the audio stream into a sequence of windows; and

b.1.2) calculating a set of 13 Mel Frequency Cepstrel Coefficients and their first and second order derivatives for each window.

8. The method of claim 1, wherein step (c) comprises executing a Viterbi algorithm.

9. The method of claim 1, wherein step (c) comprises calculating a posterior probability.

10. The method of claim 9, wherein the posterior probability comprises:

11. The method of claim 1, wherein step (c) comprises:

c.1) assigning a constant predetermined probability to the portions of the audio stream that do not match the keyword.

12. The method of claim 1, wherein step (c) comprises computing further data selected from the group consisting of:

anti-word match scores, mismatch phoneme percentage, match phoneme percentage, duration penalized probability, and a predetermined Confidence value.

13. The method of claim 12, wherein the predetermined Confidence value is chosen for each of the predetermined keywords so as to achieve a desired false alarm rate and accuracy.

14. The method of claim 1, wherein the audio stream comprises a continuous spoken speech stream.

15. A computer-implemented method for spotting predetermined keywords in an audio stream, comprising the steps of:

a) developing a keyword model for the predetermined keywords;

b) dividing the audio stream into a series of points in an acoustic space that spans all possible sounds created in a particular language;

c) computing a posterior probability that a first trajectory of each keyword model for the predetermined keywords in the acoustic space matches a second trajectory of a portion of the series of points in the acoustic space;

d) comparing the posterior probability to a predetermined threshold; and

e) reporting a spotted keyword if the posterior probability is greater than the predetermined threshold.

16. The method of claim 15, wherein step (e) comprises:

e.1) declaring a potential spotted word if the posterior probability is greater than the predetermined threshold;

e.2) computing further data to aid in determination of mismatches;

e.3) using the further data to determine if the potential spotted word is a false alarm; and

e.4) reporting spotted keyword if a false alarm is not identified at step (e.3).

17. The method of claim 15, wherein step (a) comprises concatenating phoneme hidden Markov models of predetermined keywords.

18. The method of claim 15, wherein step (a) comprises:

a.1) creating a pronunciation dictionary that defines a sequence of phonemes for each of the predetermined keywords;

a.2) creating an acoustic model that statistically models a relation between textual properties of the phonemes for each of the predetermined keywords and spoken properties of the phonemes for each of the predetermined keywords; and

a.3) concatenating acoustic models for the sequence of phonemes for each of the predetermined keywords.

19. The method of claim 18, wherein step (a.2) comprises creating a set of Gaussian mixture models.

20. The method of claim 19, wherein step (a.2) comprises creating the acoustic model selected from the group consisting of:

context-independent model, context-dependent model, and triphone model.

21. The method of claim 15, wherein step (b) comprises:

b.1) converting the audio stream into a sequence of windows; and

b.2) calculating a set of 13 Mel Frequency Cepstrel Coefficients and their first and second order derivatives for each window.

22. The method of claim 15, wherein step (c) comprises executing a Viterbi algorithm.

23. The method of claim 15, wherein the posterior probability comprises:

24. The method of claim 15, wherein step (c) comprises:

c.1) assigning a constant predetermined probability to the portions of the audio stream that do not match the keyword.

25. The method of claim 16, wherein step (e.2) comprises computing further data selected from the group consisting of:

anti-word match scores, mismatch phoneme percentage, match phoneme percentage, duration penalized probability, and a predetermined Confidence value.

26. The method of claim 25, wherein the predetermined Confidence value is chosen for each of the predetermined keywords so as to achieve a desired false alarm rate and accuracy.

27. The method of claim 15, wherein the audio stream comprises a continuous spoken speech stream.

28. The method of claim 15, wherein the space comprises a 39-dimensional space.

29. A system for spotting predetermined keywords in an audio stream, comprising:

means for developing a keyword model for the predetermined keywords;

means for comparing the keyword model and the audio stream to spot probable ones of the predetermined keywords;

means for computing a probability that a portion of the audio stream matches one of the predetermined keywords from the keyword model;

means for comparing the computed probability to a predetermined threshold;

means for declaring a potential spotted word if the computed probability is greater than the predetermined threshold;

means for computing further data to aid in determination of mismatches;

means for using the further data to determine if the potential spotted word is a false alarm; and

means for reporting spotted keyword if a false alarm is not identified.

30. The system of claim 29, wherein said means for comparing the keyword model and the audio stream to spot probable ones of the predetermined keywords is capable of pattern matching.

31. The system of claim 29, wherein said audio stream is supplied from a telephone conversation.

32. The system of claim 29, wherein said system contains a means for decomposing a word into a sequence of phonemes.

33. The system of claim 29, wherein said means for comparing the keyword model and the audio stream to spot probable ones of the predetermined keywords matches patterns between said keywords and said audio stream.