Fig. 4From: Explicit-memory multiresolution adaptive framework for speech and music separationAuditory spectrograms illustrating the workings of the self-feedback mechanism. Row 1:clean spectrogram, row 2:mixture spectrogram, row 3:output of \(L_{1+2-SI}\), row 4:output of \(TD_{1+2-SI}\), row 5:difference of \(TD_{1+2-SI}\) and \(L_{1+2-SI}\)Back to article page