摘要 |
<p>The invention concerns a method for coding speech, made up of a sequence of sound units, which consists in comparing the units to references in a library associated with primary code words, determining the differences between the units and the references, coding the differences by secondary code words to substitute the units with pairs of primary and secondary codes and, if the comparison concerns spectral energy of frequency lines, in determining, for the lines, weighting coefficients, standardized with respect to the energy of the reference lines, and integrating the coefficients in the secondary code word.</p> |