Sparse Component Analysis

Sparse Component Analysis is an underdetermined blind source separation algorithm based on an instantaneous mixing model. Previously, automated algorithms like ICA or SOBI were unavailable for this method. Our work introduces significant improvements to the original algorithm principles, enabling full automation while substantially enhancing computational efficiency and accuracy. The current implementation specifically targets vibration signal processing. For adaptation to speech signals, users need to modify the mixing matrix estimation code section (a relatively straightforward adjustment), while the core ℓ1-norm minimization component remains unchanged. The associated research paper is referenced as: Identification of modal parameters using an improved sparse blind source separation.