Code-excited iterated function prediction
Zhicheng Wang · 2002
This paper proposes a new prediction method using iterated function systems (IFS) for vector modeling of an arbitrary discrete sequence. This prediction model uses data itself for representing the future speech signal or discrete sequence and is very different from traditional methods such as LPC model or polynomial fits. On the other hand, the proposed model is different from the traditional IFS because of the backward prediction, fixed-length vector mode, and predictive residual vector usage. The model of iterated function prediction has been developed for a new speech coding scheme. The coding scheme presented is novel and unique and has great potential applications.