Low-complexity bit-robust method and system for combining codewords to form a single codeword

1. A method of combining codewords comprising:

splitting a first codeword into a most significant bits part (MSP) and a least significant bits part (LSP);

combining a MSP of the first codeword with the one or more MSPs of the one or more codewords for producing a first group;

concatenating a LSP of the first codeword with the one or more LSPs of the one or more codewords for producing a second group; and

multiplexing the first group and the second group to form the multiplexed codeword.

2. The method for claim 1 , further comprising:

de-splitting the multiplexed codeword into the first group and the second group, wherein the second group consists of the one or more LSPs;

de-combining the first group into the one or more MSPs;

concatenating the one or more MSPs with the one or more LSPs for producing the one or more codewords.

3. The method of claim 1 , wherein the concatenating is one of a scrambling or shuffling operation.

4. The method of claim 1 , further comprising precomputing a bit-length for the first group and a bit-length of the second group that reduces a computational complexity and increases a bit-error robustness of the multiplexed codeword.

5. The method of claim 4 , wherein the precomputing a bit-length for the first group comprises:

determining a number of codewords to include in the first group;

determining a number of most significant bits (MSBs) in the codewords;

computing a plurality of maximal values for the MSPs of the first group based on the MSBs;

multiplying together the plurality of maximal values for producing a product; and

determining a bit-length of the product, wherein the bit-length corresponds to the bit-length of the first group.

6. The method of claim 5 , further comprising:

selecting a bit-length for the one or more MSPs wherein the bit-length of the multiplexed codeword is equal to or less than a combined bit-length.

7. The method of claim 5 , further comprising:

selecting a bit-length for the one or more MSPs that minimizes a bit-length of the first group,

wherein a bit-length of the multiplexed codeword is the sum of the bit-length of the first group and a bit-length of the second group.

8. The method of claim 7 , wherein bit-errors in the LSP correspond to decoding errors in a codeword associated with the LSP.

9. The method of claim 1 , further comprising assigning a higher proportion of error protection to the first group than the second group.

10. The method of claim 1 , further comprising:

representing a speech signal by a sequence of pulses;

determining a plurality of pulse parameters based on the sequence of pulses;

generating one or more factorial packed codewords from the plurality of pulse parameters; and

generating the multiplexed codeword from the one or more factorial packed codewords,

wherein a factorial packed codeword has a first portion corresponding to a number of non-zero pulse positions, a second portion corresponding to a location of the non-zero pulse positions, a third portion corresponding to a number of unit magnitude pulses in each of the non-zero positions, and a fourth portion corresponding to the sign of the pulse in each of the non-zero positions.

11. The method of claim 10 , wherein the generating the one or more factorial packed codewords incorporates Huffman Coding.

12. A method for factorial packed coding comprising:

splitting a first Cπ into a Cπ most significant bits part (MSP) and a Cπ least significant bits part (LSP);

splitting a first Cμ into a Cμ MSP and a Cμ LSP;

combining the Cπ MSP and the Cμ MSP from one or more factorial packed codewords into a first group;

concatenating the Cπ LSP and the Cμ LSP with Cσ from the one or more factorial packed codewords to produce a second group; and

combining the first group and the second group with a Cν to form an error robust factorial packed codeword.

13. The method of claim 12 , further comprising selecting a bit-length of a LSP such that the minimum value of the power term over the one or more factorial packed codewords is maximized without increasing the bit-length of the combined codeword.

14. The method of claim 12 , wherein the combining further comprises:

calculating a maximal value for the Cπ MSP and the Cμ MSP for the plurality of factorial packed codewords; and

multiplying together the maximal values of the plurality of factorial packed codewords to produce a first portion of Cν.

15. The method of claim 14 , wherein the first portion of Cν is multiplied by a power term of 2 to left shift the value of Cν by a number of bits corresponding to the bit-length of the second group.

16. The method of claim 15 , wherein the power term further includes a bit-length of the Cπ LSP and a bit-length of the Cμ LSP.

17. The method of claim 12 , wherein the splitting the first Cπ and the splitting the first Cμ to produce Cν is dependent on the number of non-zero pulse positions.

18. The method of claim 12 , further comprising selecting a bit-length of a MSP to produce a bit-length of the combined codeword that is equal to the bit-length of a factorial packed codeword.

19. The method of claim 12 , further comprising a bit-length of a LSP that does not increase the bit-length of the combined codeword.

20. A system for coding a bit sequence comprising:

a splitter for splitting a first codeword into a most significant bits part (MSP) and a least significant bits part (LSP);

a combiner for combining the MSP of the first codeword with a second codeword to produce a first group; and

a concatenator for concatenating the first group with the LSP to produce a second group, and multiplexing the first group with the second group to produce a multiplexed codeword.

21. The system of claim 20 , further comprising:

an error protection encoder for adding a first error protection code to the first group and a second error protection code to the second group, wherein a higher proportion of error protection is applied to the first group than the second group; and

an error protection decoder unit for removing a first error protection code to the first group and a second error protection code to the second group, wherein a higher proportion of error protection is applied to the first group than the second group.

22. The system of claim 20 , further comprising a decoder for

de-splitting the multiplexed codeword into the first group and the second group, wherein the second group consists of one or more LSPs;

de-combining the first group into the one or more MSPs;

concatenating the one or more MSPs with the one or more LSPs to produce the one or more codewords.