METHOD FOR VOICEMAIL QUALITY DETECTION

• US 20150073785A1
• Filed: 09/06/2013
• Published: 03/12/2015
• Est. Priority Date: 09/06/2013
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Sharma, Dushyant, Naylor, Patrick
• Granted Patent: US 9,870,784 B2
• Time in Patent Office: Days
• US Class Current: 704/230
• CPC Class Codes: G10L 25/60 for measuring the quality of voice signals

Abstract

A system and method for speech quality detection is included. The method may include receiving, at a computing device, a first speech signal associated with a particular user. The method may include extracting one or more short-term features from the first speech signal wherein extracting short-term features includes extracting a time frame of between 10-50 ms. The method may also include determining one or more statistics of each of the one or more short-term features from the first speech signal. The method may further include classifying the one or more statistics as belonging to one of a set of quality classes.

Claims

1. A computer-implemented method for speech quality detection comprising:

receiving, at a computing device, a first speech signal associated with a user;

extracting one or more short-term features from the first speech signal wherein extracting short-term features includes extracting a time frame of between 10-50 ms;

determining one or more statistics of each of the one or more short-term features from the first speech signal; and

classifying the one or more statistics as belonging to one of a set of quality classes.

2. The method of claim 1, wherein the one or more statistics include at least one of mean, variance, skewness, and kurtosis.

3. The method of claim 1, wherein the one or more short-term features include at least one of linear predictive coding residual, pitch frequency, Hilbert envelope, zero crossing rate, importance weighted signal to noise ratio, and difference from long-term average speech magnitude spectrum features.

4. The method of claim 1, wherein classifying is based upon, at least in part, non-intrusive classification of message quality.

5. The method of claim 4, wherein classifying is performed per each time frame.

6. The method of claim 1 wherein classifying the one or more statistics includes modeling a speech quality class using a binary tree classifier.

7. The method of claim 3, wherein the difference from long-term average speech magnitude spectrum features includes at least one of flatness, centroid, and a power spectrum of long term deviation.

8. The method of claim 1, further comprising:

extracting one or more long-term features from the first speech signal.

9. The method of claim 8, wherein the one or more long-term features includes a percentage of energy per frequency band.

10. The method of claim 1, further comprising:

automatically generating at least one training database, based upon, at least in part, the first speech signal and an intrusive speech quality algorithm.

11. A non-transitory computer-readable storage medium having stored thereon instructions, which when executed by a processor result in one or more operations, the operations comprising:

receiving, at a computing device, a first speech signal associated with a particular user;

extracting one or more short-term features from the first speech signal wherein extracting short-term features includes extracting a time frame of between 10-50 ms;

determining one or more statistics of each of the one or more short-term features from the first speech signal; and

classifying the one or more statistics as belonging to one of a set of quality classes.

12. The non-transitory computer-readable medium of claim 11, wherein the one or more statistics include at least one of mean, variance, skewness, and kurtosis.

13. The non-transitory computer-readable medium of claim 11, wherein the one or more short-term features include at least one of linear predictive coding residual, pitch frequency, Hilbert envelope, zero crossing rate, importance weighted signal to noise ratio, and difference from long-term average speech magnitude spectrum features.

14. The non-transitory computer-readable medium of claim 11, wherein classifying is based upon, at least in part, non-intrusive classification of message quality.

15. The non-transitory computer-readable medium of claim 14, wherein classifying is performed per each time frame.

16. The non-transitory computer-readable medium of claim 11, wherein classifying the one or more statistics includes modeling a speech quality class using a binary tree classifier.

17. A system comprising:

one or more processors configured to receive a first speech signal associated with a particular user, the one or more processors further configured to extract one or more short-term features from the first speech signal wherein extracting short-term features includes extracting a time frame of between 10-50 ms, the one or more processors further configured to determine one or more statistics of each of the one or more short-term features from the first speech signal, the one or more processors further configured to classify the one or more statistics as belonging to one of a set of quality classes.

18. The system of claim 17, wherein the one or more statistics include at least one of mean, variance, skewness, and kurtosis.

19. The system of claim 17, wherein the one or more short-term features include at least one of linear predictive coding residual, pitch frequency, Hilbert envelope, zero crossing rate, importance weighted signal to noise ratio, and difference from long-term average speech magnitude spectrum features.

20. The system of claim 17, wherein classifying is based upon, at least in part, non-intrusive classification of speech quality.