Removing noise from feature vectors

1. A method of identifying a clean signal feature vector from a noisy signal feature vector, the method comprising:

generating at least two mixture components for a prior probability describing combinations of clean signal feature vectors with obscuring feature vectors, each mixture component being generated by combining at least one distribution of obscuring feature vectors that forms part of a mixture of distributions that describes a prior probability of the obscuring feature vectors with a distribution of clean signal feature vectors that forms part of a mixture of distributions that describes a prior probability of clean signal feature vectors such that a mean for a mixture component formed by the combination comprises a mean for the distribution of obscuring feature vectors and a mean for the distribution of clean signal feature vectors wherein at least one obscuring feature vector is a channel distortion feature vector associated with a first channel and at least one other obscuring feature vector is a channel distortion feature vector associated with a second channel; and

using each mixture component of the prior probability and the noisy signal feature vector to identify the clean signal feature vector.

2. The method of claim 1 wherein at least one obscuring feature vector is a noise feature vector.

3. The method of claim 2 wherein generating at least two mixture components comprises generating a separate mixture component for each combination of a distribution of noise feature vectors with a distribution of clean signal feature vectors.

4. The method of claim 1 wherein at least one obscuring feature vector is a channel distortion feature vector.

5. The method of claim 4 wherein generating at least two mixture components comprises generating a separate mixture component for each combination of a distribution of channel distortion feature vectors with a distribution of clean signal feature vectors.

6. The method of claim 1 wherein the clean signal feature vectors comprise clean signal feature vectors from at least two sources.

7. The method of claim 1 wherein identifying the clean signal feature vector comprises using algorithms obtained through an approximate Bayesian inference technique to identify the clean feature vectors.

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

receiving a feature vector representing a portion of a noisy signal; and

identifying a feature vector representing a portion of a clean signal from the feature vector for the noisy signal through steps comprising:

combining at least two distributions of obscuring feature vectors, wherein each distribution of obscuring feature vectors forms part of a separate mixture of distributions of obscuring feature vectors, with at least one distribution of model clean signal feature vectors to form a distribution that forms part of a mixture of distributions that describe a prior probability of combinations of obscuring feature vectors and cleans signal feature vectors wherein one of the distributions of obscuring feature vectors comprises a distribution of model channel distortion feature vectors associated with a first channel and another of the distributions of obscuring feature vectors comprises a distribution of channel distortion feature vectors associated with a second channel that is different from the first channel; and

using the mixture of distributions of the prior probability and the feature vector for the noisy signal to identify the feature vector for the clean signal.

9. The computer-readable storage medium of claim 8 wherein the obscuring feature vectors are model noise feature vectors.

10. The computer-readable storage medium of claim 8 wherein the at least one distribution of model clean signal feature vectors comprises at least one model clean signal feature vector from a first source and at least one model clean signal feature vector from a second source.