A digital filter design and optimization method for sampling rate alteratio
n of narrow-band signals is presented. The resulting predictive finite-dura
tion impulse-response filters are used to remove the aliasing or imaging fr
equency components when the sampling rate is changed. The sampling rate can
be changed either by an integer factor or a rational factor. A least-squar
es-based generalization is developed for optimal wide-band noise attenuatio
n.