Speech modeling and enhancement based on magnitude-normalized spectra

• US 20070150263A1
• Filed: 12/23/2005
• Published: 06/28/2007
• Est. Priority Date: 12/23/2005
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Zhang, Zhengyou, Acero, Alejandro, Liu, Zicheng, Subramanya, Amarnag
• Granted Patent: US 7,930,178 B2
• Time in Patent Office: Days
• US Class Current: 704/205
• CPC Class Codes: G10L 15/20 Speech recognition techniques specially adapted for robustness in adverse environments, e.g. in noise, of stress induced speech G10L21/02 takes precedence
• CPC Class Codes: G10L 21/0208 Noise filtering

Abstract

A frame of a speech signal is converted into the spectral domain to identify a plurality of frequency components and an energy value for the frame is determined. The plurality of frequency components is divided by the energy value for the frame to form energy-normalized frequency components. A model is then constructed from the energy-normalized frequency components and can be used for speech recognition and speech enhancement.

Claims

1. A method comprising:

converting a frame of a speech signal into the spectral domain to identify a plurality of frequency components;

determining an energy value for the frame;

dividing the plurality of frequency components of the speech signal by the energy value for the frame to form energy-normalized frequency components; and

constructing a model from the energy-normalized frequency components.

2. The method of claim 1 wherein constructing the model comprises clustering frames of energy-normalized frequency components into mixture components and forming separate model parameters for each mixture component.

3. The method of claim 2 wherein clustering frames comprises determining the difference between a log magnitude of an energy-normalized frequency component in a first frame and a log magnitude of an energy-normalized frequency component in a second frame.

4. The method of claim 1 further comprising filtering a D.C. frequency component from the plurality of frequency components to form filtered frequency components before determining the energy value.

5. The method of claim 4 wherein dividing the plurality of frequency components by the energy value comprises dividing the filtered frequency components by the energy value.

6. The method of claim 1 further comprising using the model to determine a likelihood for a frame of an input speech signal by:

converting the frame of the input speech signal to the spectral domain to produce a plurality of frequency components;

determining an energy value for the frame of the input signal;

dividing the plurality of frequency components of the frame of the input speech signal by the energy value for the frame of the input signal to form input energy-normalized frequency components; and

applying the input energy-normalized frequency components to the model to determine the likelihood.

7. The method of claim 1 further comprising using the model to estimate a clean speech value from a noisy speech signal.

8. A computer-readable medium having computer-executable instructions for performing steps comprising:

receiving values representing a noisy speech signal; and

using a model of energy-normalized clean-speech spectral values to estimate a noise-reduced value from the noisy speech signal.

9. The computer-readable medium of claim 8 wherein using the model of energy-normalized clean speech spectral values comprises using a mixture model of energy-normalized clean-speech spectral values.

10. The computer-readable medium of claim 8 wherein estimating a noise-reduced value further comprises estimating a gain value that describes the ratio of clean speech spectral values to energy-normalized clean speech spectral values.

11. The computer-readable medium of claim 10 wherein estimating a noise-reduced value comprises estimating energy-normalized noise-reduced speech spectral values and multiplying the energy-normalized noise-reduced speech spectral values by the gain value to produce noise-reduced values.

12. The computer-readable medium of claim 11 further comprising estimating a separate gain value for each of a plurality of frames of the noisy speech signal.

13. The computer-readable medium of claim 8 wherein estimating a noise-reduced value further comprises utilizing a model of a speech state that provides the probability that a frame of the noisy speech signal contains speech.

14. The computer-readable medium of claim 8 wherein estimating a noise-reduced value comprises estimating the noise-reduced value based on an alternative sensor signal.

15. A method comprising:

receiving an air conduction microphone signal;

receiving an alternative sensor signal;

using the air conduction microphone signal, the alternative sensor signal, and a model of energy-normalized clean speech spectral values to estimate a noise-reduced speech value.

16. The method of claim 15 wherein estimating a noise-reduced speech value comprises estimating energy-normalized noise-reduced speech spectral values and converting the energy-normalized noise-reduced speech spectral values into noise-reduced speech values.

17. The method of claim 16 wherein converting the energy-normalized noise-reduced speech spectral values comprises multiplying the energy-normalized noise-reduced speech spectral values by a gain value.

18. The method of claim 17 further comprising iterating between estimating energy-normalized noise-reduced speech spectral values and estimating the gain value.

19. The method of claim 17 wherein estimating the gain value comprises estimating a separate gain value for each frame of the air conduction microphone signal.

20. The method of claim 15 wherein the model of energy-normalized clean speech spectral values comprises a mixture model.