High-quality encoding at low-bit rates

1. A method comprising:

splitting, at an apparatus, an input vector into an outlier portion and a stationary portion;

encoding the outlier portion of the input vector;

dividing the stationary portion into one or more subvectors;

classifying each of said subvectors as trivial or non-trivial;

encoding each said trivial subvector using pre-defined sample values; and

encoding each said non-trivial subvector with at least one location of at least one significant sample and a sign of said at least one significant sample.

2. The method of claim 1 wherein the steps of the method are implemented as computer-executable instructions stored on a computer readable medium.

3. The method of claim 1 wherein the encoding of the outlier portion comprises encoding amplitudes, signs and positions of samples of the outlier portion.

4. The method of claim 3 wherein the encoding of the amplitudes of the samples of the outlier portion comprises using a Golomb-Rice code.

5. A method comprising:

determining whether at least one outlier sample is present within a data vector;

decoding, at an apparatus, said at least one outlier sample by retrieving information relative to magnitude, sign, and position of the outlier sample;

processing information about one or more subvectors within the data vector;

processing encoded data for at least one subvector at said apparatus in order to identify information including a binary codeword and sign information for said at least one subvector; and

using the information in order to perform reconstruction.

6. The method of claim 5 wherein the steps of the method are implemented as computer-executable instructions stored on a computer readable medium.

7. The method of claim 5 wherein the decoding of said at least one outlier sample comprises using a Golomb-Rice code.

8. The method of claim 5 wherein the processing of the encoded data for said at least one subvector comprises decoding a codeword of a binary Golay code.

9. A method comprising:

encoding, at an apparatus, a vector with at least one location of a significant sample and a sign of the significant sample, said location of the significant sample within the vector is encoded using a nearest neighbor in a selected portion of binary code;

encoding a second vector with at least one location of a significant sample and a sign of a significant sample,

wherein said location of a significant sample within said vector is encoded using a nearest neighbor in a first selected portion of binary code and said location of a significant sample within said second vector is encoded using a nearest neighbor in a second selected portion of binary code, and wherein said first and second locations are jointly encoded to minimize distortion.

10. The method of claim 9 wherein the steps of the method are implemented as computer-executable instructions stored on a computer readable medium.

11. The method of claim 9 wherein said selected portion of a binary code comprises a shell of the binary code having a predetermined Hamming weight.

12. The method of claim 9 wherein the binary code is a binary Golay code.

13. The method of claim 9 wherein said first selected portion of binary code is different from said second selected portion of binary code.

14. An apparatus comprising:

a processor for executing computer executable instructions;

memory that stores the computer executable instructions;

wherein the computer executable instructions cause the apparatus to:

split an input vector into an outlier portion and a stationary portion;

encode the outlier portion of the input vector;

divide the stationary portion into one or more subvectors;

classify each of said subvectors as trivial or non-trivial;

encode each said trivial subvector using pre-defined sample values; and

encode each said non-trivial subvector with at least one location of at least one significant sample and a sign of said at least one significant sample.

15. The apparatus of claim 14 wherein the computer executable instructions cause the apparatus to:

encode amplitudes, signs and positions of samples of the outlier portion.

16. The apparatus of claim 15 wherein the computer executable instructions cause the apparatus to:

encode the amplitudes of the samples of the outlier portion using at least a Golomb-Rice code.

17. An apparatus comprising:

a processor for executing computer executable instructions;

memory that stores the computer executable instructions;

wherein the computer executable instructions cause the apparatus to:

determine whether at least one outlier sample is present within a data vector;

decode said at least one outlier sample by retrieving information relative to magnitude, sign, and position of the outlier sample;

process information about one or more subvectors within the data vector;

process encoded data for at least one subvector in order to identify information including a binary codeword and sign information for said at least one subvector; and

use the information in order to perform reconstruction.

18. The apparatus of claim 17 wherein the computer executable instructions cause the apparatus to:

decode a codeword of a binary Golay code when processing the encoded data for said at least one subvector.

19. An apparatus comprising:

a processor for executing computer executable instructions;

memory that stores the computer executable instructions;

wherein the computer executable instructions cause the apparatus to:

encode a vector with at least one location of a significant sample and a sign of the significant sample, said location of the significant sample within the vector is encoded using a nearest neighbor in a selected portion of binary code;

encode a second vector with at least one location of a significant sample and a sign of a significant sample,

wherein said location of a significant sample within said vector is encoded using a nearest neighbor in a first selected portion of binary code and said location of a significant sample within said second vector is encoded using a nearest neighbor in a second selected portion of binary code, and wherein said first and second locations are jointly encoded to minimize distortion.