Acoustic signal processing is one of the earliest fields in which the source separation problem was studied. Indeed, most available acoustic signals are mixtures of several sources. Although good separation may currently be obtained for some simple synthetic mixtures, the separation of real-world signals requires basic competence in this field and remains difficult to achieve. It should therefore be stressed that acoustics is among the most difficult application fields of source separation under investigation. Early acoustic source separation systems relied on fixed or adaptive beamforming, which remains in use today. These systems require some prior knowledge, such as the relative positions of the microphones and the target source or the time intervals during which the target source is inactive. In practice, prior knowledge about the sources or the mixing system is rarely available, so that blind source separation (BSS) systems must be used instead. In this chapter, we discuss the application of the BSS techniques reviewed in this book to the separation of audio signals, with main emphasis on convolutive independent component analysis (ICA) (see Chapter 8) and sparse component analysis (SCA) (see Chapter 10). In particular, we characterize real-world audio mixtures and summarize the performance of existing systems for such mixtures.