Implementation of a 4.8 kbps voice codec based on pitch-synchronous DFT coding

A 4.8 kb/s voice codec based on a coding algorithm using pitch-synchronous DFT (discrete Fourier transform) is described. This algorithm combines waveform repetition with pitch-synchronous DFT spectral coding in which the phase information and nonharmonic spectral components are ignored. This implementation introduces adaptive postspectral emphasis for the enhancement of the reproduced speech quality. The prototype codec uses three Fujitsu MB86232 digital signal processor (DSP) chips. The reproduced speech signal quality was evaluated, and the performance of adaptive postspectral emphasis was compared with that of adaptive postfiltering. The reproduced speech was found to be good. Pitch-synchronous DFT thus expresses the power spectra well. Adaptive postspectral emphasis was shown to be as effective as adaptive postfiltering for enhancing the output signal quality. This does not add complexity to frequency-domain coding