Method of determining uncertainty associated with acoustic distortion-based noise reduction

• US 7,289,955 B2
• Filed: 12/20/2006
• Issued: 10/30/2007
• Est. Priority Date: 05/20/2002
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Deng, Li, Droppo, James G., Acero, Alejandro
• Primary Examiner(s): Lerner; Martin
• Application Number:
  US11/642,389
• Publication Number:
  US 20070106504A1
• Time in Patent Office: 314 Days
• Field of Search: 704/226, 704/227, 704/228, 704/233, 704/236, 704/240, 704/243, 704/244
• US Class Current: 704/226
• 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 method and apparatus are provided for determining uncertainty in noise reduction based on a parametric model of speech distortion. The method is first used to reduce noise in a noisy signal. In particular, noise is reduced from a representation of a portion of a noisy signal to produce a representation of a cleaned signal by utilizing an acoustic environment model. The uncertainty associated with the noise reduction process is then computed. In one embodiment, the uncertainty of the noise reduction process is used, in conjunction with the noise-reduced signal, to decode a pattern state.

Claims

1. A method comprising:

reducing noise in a representation of a portion of a noisy signal to produce a representation of a portion of a noise-reduced signal using an acoustic-environment model;

identifying an uncertainty associated with reducing the noise by computing a difference between an expectation for the square of the portion of the noise-reduced signal and the square of the expectation for the portion of the noise-reduced signal; and

using the uncertainty and the portion of the noise-reduced signal to decode a pattern state.

2. The method of claim 1 using the uncertainty and the portion of the noise-reduced signal to decode a pattern state comprises:

using the uncertainty to adjust a probability distribution to form a modified probability distribution; and

using the representation of a portion of the noise-reduced signal and the modified probability distribution to decode the pattern state.

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

converting a frame of a noisy signal into a feature vector comprising at least two components;

reducing noise in a component of the feature vector for the noisy signal to produce a component of a feature vector for a cleaned signal by utilizing an acoustic distortion model;

identifying an uncertainty associated with reducing the noise from the component by computing a difference between an expectation for a square of a feature vector for a cleaned signal and a square of the expectation for a feature vector for a cleaned signal; and

using the uncertainty to decode a phonetic state.

4. The computer-readable medium of claim 3 wherein the computer-executable instructions perform further steps comprising:

determining a probability component of a probability for a phonetic state based in part on the component for the cleaned signal and the uncertainty associated with reducing the noise from the component; and

using the probability component to determine the probability of the phonetic state regardless of the value of the uncertainty.

5. The computer-readable medium of claim 4 wherein determining a probability component comprises defining a probability distribution based in part on the uncertainty.

6. The computer-readable medium of claim 5 wherein defining a probability distribution comprises adding the uncertainty to a variance of a probability distribution.