Codebook tables for encoding and decoding

A variable rate speech compression system for processing a speech signal, the variable rate speech compression system comprising:

• an encoding system operable to determine a bit rate and a type classification for the speech signal, the bit rate comprising a first rate and a second rate, and the type classification comprising a first type and a second type, the encoding system comprising:
• a line spectrum frequency prediction error quantization table selectable as a function of the bit rate, wherein the encoding system is operable to encode short-term predictor parameters of the speech signal with the line spectrum frequency prediction error quantization table;
• an interpolation module operable with the line spectrum frequency prediction error quantization table to encode short-term predictor parameters, when the bit rate is the first rate and the type classification is the first type;
• a line spectrum frequency predictor coefficient table selectable as a function of the bit rate, wherein the encoding system is operable to generate predicted line spectrum frequencies with the line spectrum frequency predictor coefficient table;
• and
• a predictor switch module operable with the line spectrum frequency predictor coefficient table to generate predicted line spectrum frequencies, when the bit rate is the second rate.

A method of processing a speech signal with a variable rate speech compression system, the method comprising:

• determining a rate and a type for the speech signal;
• encoding short-term predictor parameters of the speech signal with a line spectrum frequency prediction error quantization table as a function of the rate;
• jointly encoding an adaptive codebook gain and a fixed codebook gain of the speech signal with a 2D gain quantization table when the type is a first type;
• encoding the adaptive codebook gain with a pre-gain quantization table as a function of the rate when the type is a second type;
• and
• encoding the fixed codebook gain with a delayed gain quantization table as a function of the rate when the type is the second type.

A method of processing a speech signal, the method comprising:

• selecting a bit rate and a type classification;
• converting short-term predictor parameters extracted from the speech signal to line spectrum frequencies;
• determining predicted line spectrum frequencies with a line spectrum frequency predictor coefficients table when the bit rate selected is a first rate;
• determining predicted line spectrum frequencies with a line spectrum frequency predictor coefficients table and a predictor switch module when the bit rate selected is a second rate;
• subtracting predicted line spectrum frequencies from line spectrum frequencies to generate a line spectrum frequencies prediction error;
• quantizing the line spectrum frequencies predication error to produce quantized line spectrum frequencies;
• and
• modifying the quantized line spectrum frequencies with an interpolation module when the bit rate selected is the first rate and the type classification is a first type;
• wherein when the bit rate is the second rate, determining predicted line spectrum frequencies comprises selecting one of:
• a first set of predictor coefficients, the first set of predictor coefficients including a first vector represented as {0.45782564, 0.59002827, 0.73704688, 0.73388197, 0.75903791, 0.74076479, 0.65966007, 0.58070788, 0.52280647, 0.42738207};
• and
• a second set of predictor coefficients, the second set of predictor coefficients including a first vector represented as {0.14936742, 0.25397094, 0.42536339, 0.40318214, 0.39778242, 0.34731435, 0.22773174, 0.17583478, 0.12497067, 0.11001108}.