This text provides an overview of the various approaches used for noise reduction, with an emphasis on the differents signal representations. In particular, the classic methodologies for the individuation and removal of local and global disturbances are described. Moreover, some innovative algorithms are illustrated in detail: the algorithms based in the frequency domain are explained, as well as those based in the time domain. It is very difficult to identify evaluation criteria for the results of the various audio restoration algorithms. The last section focus on some numeric indexes, which aim at indicating the effects of an audio restoration process.