Data reduction of audio signals may be employed to optimize the use of storage capacity in digital recording systems, or to make cost-effective use of bandwidth in broadcast transmission systems and in digital communications. Nothing is inherently wrong with linear PCM from a sound quality point of view; indeed, it is probably the best thing to use, all else being equal. Speech coders are specifically designed for speech communication systems, where intelligibility can be more important than high fidelity, and these can be designed to operate at extremely low bit rates. Audio data reduction falls into the two distinct categories of lossless and lossy coding. One variant on lossy low-bit-rate coding involves the encoding of audio signals in the form of a core signal alongside a sparse stream of parameters that describe one or more features needed to reconstruct an approximation of the original signals.