A design of 0.9kb/s speech coding algorithm based on MELPe

A 0.9kb/s speech coder based on the 2.4kb/s Enhanced Mixed Excitation Liner Prediction (MELPe) analysis algorithm is presented in this paper. At the part of speech coding, the feature parameters of three consecutive frames are extracted using the same algorithm as in the 2.4kb/s MELPe standard and then grouped into a super-frame, new quantization schemes are designed to take advantage of the long block size and save bits by using vector quantization, joint quantization and an interpolation method. At the part of the speech decoding, all the synthesis parameters are interpolated pitch-synchronously for each synthesized pitch period with a new interpolated factor, and a modified harmonic synthesis algorithm is used to generate the mixed excitation. Subjective test results indicate that the 0.9kb/s speech coder also obtains a high quality which is slightly worse than the proposed federal standard 2.4kb/s Mixed Excitation Liner Prediction (MELP) coder.