Method and an apparatus for voice quality enhancement (VQE) for detection of VQE in a receiving signal using a guassian mixture model

Patent metadata

Field	Value
Publication number	US-9299359-B2
Application number	US-201313940494-A
Country	US
Kind code	B2
Filing date	Jul 12, 2013
Priority date	Jan 14, 2011
Publication date	Mar 29, 2016
Grant date	Mar 29, 2016

Abstract

A voice quality enhancement (VQE) detector for a network element receiving an audio signal from a previous network element of a network, wherein said voice quality enhancement detector is adapted: to perform a voice quality enhancement detection based on the received audio signal, wherein said voice quality enhancement detection comprises detecting that at least one voice quality enhancement function, VQEE comprising of a noise cancellation or an echo cancellation function using a Gaussian Mixture Model was applied to the received audio signal by at least one previous network element of the network; and to control a voice quality enhancement processing of the received audio signal depending on the detection result.

First claim

What is claimed is:

1. A voice quality enhancement (VQE) detector device for a network element receiving an audio signal from a previous network element of a network,
wherein the VQE detector device is configured to:
perform a VQE detection based on the received audio signal, wherein the VOE detection comprises:
extracting and analyzing signal features from the received audio signal based on a statistical model, to detect that at least one voice quality enhancement function (VQEF) has been applied to the received audio signal by at least one previous network element of the network, wherein the statistical model is built on a database consisting of signals on which at least one type of VQE has been applied; and
control a VQE processing of the received audio signal according to a detection result;
wherein the VQEF comprises a noise reduction or an echo cancellation function;
and the statistical model is formed by a Gaussian Mixture Model GMM.

2. The VQE detector device according to claim 1 ,
wherein the at least one VQEF is at least one of the following: a noise reduction function, an echo cancellation function, a dynamic range compression function, and an automatic gain control function.

3. The VQE detector device according to claim 2 ,
wherein the VQE detector device is configured to extract and analyze signal features from the received audio signal to perform the voice quality enhancement detection, wherein the received audio signal is coded or non-coded.

4. The VQE detector device according to claim 3 ,
wherein the signal extracted and analyzed features of the received audio signal comprise at least one of the following features:
a spectral flatness measure,
a spectral dynamic,
a pitch period,
a pitch gain, and
a speech variance of the received audio signal.

5. The VQE detector device according to claim 4 ,
wherein the VQE detector device is configured to detect a strength of a VQEF applied to the received audio signal by the at least one of the previous network element.

6. The VQE detector device according to claim 4 ,
wherein the detection result provided by the VQE detector device is signaled to at least one other network element of the network.

7. The VQE detector device according to claim 2 ,
wherein said received audio signal is a coded audio signal; and
wherein said VQE detector device is connected to a signal input and processes a bit stream of the coded audio signal received at the signal input to perform said voice quality enhancement detection; or
wherein said VQE detector device is connected to an output of a bit stream decoder and processes a decoded signal generated by said bit stream de-coder based on the received coded audio signal to perform said voice quality enhancement detection.

8. The VQE detector device according to claim 1 , wherein the voice quality enhancement detection is a closed loop voice quality enhancement detection and comprises:
applying a VQEF to the received audio signal by a VQE device to obtain a processed version of the received audio signal; and
detecting that the at least one VQEF or at least the applied VQEF has been applied to the received audio signal by one of the previous network elements, if an evaluation of a voice quality metric obtained for the received audio signal and a voice quality metric obtained for the processed version of the received audio signal fulfills a predetermined criterion.

9. The VQE detector device according to claim 8 ,
wherein the VQE is applied in a closed loop where mean opinion scores between the received audio signal and a VQE processed signal are compared.

10. The VQE detector device according to claim 1 ,
wherein the VQE detector device is configured to extract and analyze signal features from the received audio signal to perform the voice quality enhancement detection, wherein the received audio signal is coded or non-coded.

11. The VQE detector device according to claim 10 ,
wherein the signal extracted and analyzed features of the received audio signal comprise at least one of the following features:
a spectral flatness measure,
a spectral dynamic,
a pitch period,
a pitch gain, and
a speech variance of the received audio signal.

12. The VQE detector device according to claim 1 ,
wherein received audio signal is a coded audio signal; and
wherein the VQE detector device is connected to one of a signal input and an output of a bit stream decoder;
when the VQE detector device is connected to the signal input, the VQE detector device processes a bit stream of the coded audio signal received at the signal input to perform the voice quality enhancement detection; and
when the VQE detector device is connected to the output of the bit stream decoder, and the VQE detector device processes a decoded signal generated by the bit stream de-coder based on the received coded audio signal to perform the voice quality enhancement detection.

13. The VQE detector device according to claim 1 ,
wherein said voice quality enhancement processing is an adaptive voice quality enhancement processing and said VQE detector device adjusts parameters of said adaptive voice quality enhancement processing depending on the detection result.

14. The VQE detector device according to claim 1 ,
wherein the VQE detector device controls to switch the at least one VQEF provided by the voice quality enhancement processing on or off according to the detection result.

15. The VQE detector device according to claim 1 ,
wherein the VQE detector device comprises a non-intrusive speech quality assessment of the received audio signal.

16. The VQE detector device according to claim 1 ,
wherein the VQE detector device is configured to detect a strength of a VQEF applied to the received audio signal by the at least one previous network element.

17. The VQE detector device according to claim 1 ,
wherein the detection result provided by the VQE detector device is signaled to at least one other network element of the network.

18. A network element of a wired and/or wireless communication network comprising
a voice quality enhancement (VQE) detector for the network element receiving an audio signal from a previous network element of the network;
at least one VQE device for the VQE processing of the received audio signal controlled by the VQE detector;
wherein the VQE detector is configured to:
perform a VQE detection based on the received audio signal, wherein the VOE detection comprises:
extracting and analyzing signal features from the received audio signal based on a statistical model, to detect that at least one voice quality enhancement function (VQEF) has been applied to the received audio signal by at least one previous network element of the network, wherein the statistical model is built on a database consisting of signals on which at least one type of VQE has been applied;
and control a VQE processing of the received audio signal according to a detection result;
wherein the VQEF comprises of a noise reduction or an echo cancellation function;
and the statistical model is formed by a Gaussian Mixture Model GMM.

19. The network element according to claim 18 , comprising:
a user terminal or a network node within a communication chain of an optical, electrical, electromagnetic wired or wireless communication network between user terminals.

20. A method for controlling a voice quality enhancement (VQE) provided by a VQE device of a network element receiving an audio signal from a previous network element

Assignees

• Huawei Tech Co Ltd

Inventors

• Taleb Anisse
• Virette David
• Xu Jianfeng

Classifications

• G10L21/02 Primary

  Speech enhancement, e.g. noise reduction or echo cancellation (reducing echo effects in line transmission systems H04B3/20; echo suppression in hands-free telephones H04M9/08) · CPC title

• H04Q3/00

  Selecting arrangements (H04Q5/00 - H04Q11/00 take precedence) · CPC title

• G10L25/69

  for evaluating synthetic or decoded voice signals · CPC title

• G10L21/0208 Primary

  Noise filtering · CPC title