An improved sparse component analysis (SCA) is developped. The SCA method is just defined in a framework before, but there no existing complete algorithm. We explore a compelte and automatical algorithm, then use it to deal with modal identification issue in machnical engineering. This software is just suitable for vibration signals, not for speech signal. If you want to process speech signals, you need to change the mixing matrix estimation method.
Dear Pro. Ishwarya Venkatesh, the corresponding paper has been submitted to shock and vibration.
Due to SCA based on instaneous mixing model, so it is only able to process sensor data without time-delay. So I advice you processing the data recorded in rigid structure instead of flexible structure.