“…An attractive reason for using sub-band coding is that potentially it can be implemented very efficiently using recent filter technologies such as CCD's or digital filters. By using long time prediction, i.e., pitch prediction, as well as short-time prediction time domain coding techniques can achieve further improvements in performance [7,8]. Similarly, it has recently been shown [9] that, by using large blocks, block transform techniques can also exploit pitch structure, as well as formant structure, entirely in the frequency domain and that this approach also leads to a corresponding improvement in performance over non-pitch-predicting methods.…”