Wideband speech coding using ADPCM and a new spectral replication method based on parametric stereo coding

A spectral replication coding scheme based on parametric stereo is introduced in this paper. Parametric stereo technique was initiated to encode both the left and right stereo channels more efficiently by extracting redundancy between these channels. A mono downmix signal plus small side information (some parameters) are transmitted instead of two channels. In the decoder both channels are reconstructed from the mono downmix signal and the received parameters. We have borrowed this technique to create a new spectral replication coding. In the proposed method, the signal is divided into two low and high frequency parts. We deal, then, with these parts as left and right channels in the stereo signal to acquire the parameters where the low part signal plays the role of the downmix signal which is coded using ADPCM technique. The results show that we can achieve 16.6-21.5% reduction in bit rate, in comparison with G.722 standard, bearing little degradation in quality.