When we listen to a song, we can distinguish the singer's voice, the guitar, the drums or other accompanying instruments in the song with little effort. When we hear a conversation in a noisy background on the street, with other people's competing conversation nearby or with a background music, we can still make a lot of sense of the conversation we focus on. This is not easily done in a computer. After all, anything that has been mixed is hard to separate. The aim is then to separate or extract each sound component(to be defined), given a mixture(or mixtures) of the signal recorded.