Bernd Iser and Gerhard Schmidt

Harman/Becker Automotive Systems, Ulm, Germany

In this chapter an introduction on bandwidth extension of telephony speech is given. Why current telephone networks apply a limiting bandpass, what kind of bandpass is used, and what can be done to (re)increase the bandwidth on the receiver side without changing the transmission system is presented. Therefore, several approaches—most of them based on the source-filter model for speech generation—are discussed. The task of bandwidth extension algorithms that make use of this model can be divided into two subtasks: an excitation signal extension part and a wideband envelope estimation part. Different methods for accomplishing these tasks, like nonlinear processing, the use of signal and noise generators, or modulation approaches on the one hand and codebook approaches, linear mapping schemes or neural networks on the other, are presented.


Speech is the most natural and convenient way of human communication. This is the reason for the big success of the telephone system since its invention in the nineteenth century [1]. At that time people didn't think of high quality speech but nowadays this has changed. Often customers are not satisfied with the quality of service provided by the telephone system especially when compared to other audio sources, such as radio or compact disc. The degradation of speech quality using analog telephone systems is

Get Adaptive Signal Processing: Next Generation Solutions now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.