Used in Dolby AC-2 and MPEG codecs, whereby an input signal is analyzed within the frequency domain as a series of narrow bands,; e.g., an AC-2 encoder/decoder uses 256 narrow bands. Frequency-masked bands whose information is inaudible because of sustained tones or transients will be ignored by such systems, thereby enabling the compression algorithm to concentrate its bit allocation on those bands which contain subjectively relevant information. In addition, transform-based systems use a system of predefined waveform patterns from an established “wavelook-up” tablelibrary of sound models and send only the identity code to the decoder to “resynthesize” the closest-fit library model. As opposed to, for example, ADPCM.