In this article, we provided an overview of some of the current linear predictive vocoder methods for speech and audio coding. Speech coding research has come a long way since the early 1990s, and several algorithms are rapidly finding their way in consumer products ranging from wireless cellular telephones to computer multimedia systems. Research and development in code excited linear prediction yielded several algorithms that have been adopted, or are strong candidates for adoption, in international wired and wireless communication standards from 16 down to 4.8 kbps. On the other hand, research activities are concentrating on the development of toll-quality algorithms that operate under 4 kbps. ITU has called for proposals for toll quality at 4 kbps for use in future videophones, personal communications, and third-generation cellular systems. Several proposals have been submitted [45-52] with none selected as of yet. Our coverage focussed on vocoder methods based on linear prediction because they have been the most popular in more recent low-rate communications and media standards. We must mention that in addition to the linear predictive coders, some frequency-domain methodologies have also been used in low-rate speech coding. In particular, sinusoidal analysis-synthesis schemes have been used as alternative to LPC. The sinusoidal model [8] represents speech by a linear combination of sinusoids. A low-rate sinusoidal transform coder (STC) has been considered in federal and TIA standardization competitions and is currently considered for scalable wideband and high-fidelity applications. Another sinusoidal system, the improved multi-band excitation (IMBE) coder [9,10], became part of the Australian (AUSSAT) mobile satellite standard and the International Mobile Satellite (INMARSAT-M) standard. [ABSTRACT FROM AUTHOR]