I'm looking for an existing implementation for a three-components decomposition of audio signals.
The idea of separate a signal into three components, one for the Sinusoidal part, one for the transients and one for the Residual (these models are often called STN models), is described in "A Review on Techniques for the Extraction of Transients in Musical Signals" by Laurent Daudet. Related methods, mentioned in this paper, include the "Time-Frequency Jigsaw Puzzle", the "Matching Pursuit" and the "Fast Iterated Reweighted Sparsifier".
Does anyone know of any functions or existing code related to this topic?
Thanks for your help.