CPSP is a signal analysis and signal visualization tool for sound, vibration or any other similar signal. CPSP stands for Continuity Preserving Signal Processing. Key property of CPSP is the use of a signal processing device that, like our ears, allows an optimal representation of all continuous developments in sounds. The preservation of this continuity is a key property of our hearing because it is used to track the development of sound sources. Well known signal processing techniques such as the Fast Fourier Transform or the Discrete Wavelet Transform do not preserve continuity well enough to approach the human performance. So the principles of CPSP are in auditory modeling and perception research, but although CPSP uses a very efficient numerical model of the human inner ear, it is in the first place a tool for signal processing, that might be used as a model of the human hearing system.

Signal analysis is all about deriving correct conclusions from a signal. In many interesting situations the signal originates from a number of physical processes. In these cases it might not be trivial to separate the signal into evidence about each of the processes. This is especially true if all or some or the processes are unknown. Our auditory system is the best known analysis system for arbitrary, and possibly even unknown, sounds. Somehow we are seldom confronted with a situation in which we mix up information of multiple sound sources and, as a consequence, reach a completely wrong conclusion. Of course our auditory system can rely on many years of experience, but an essential part is the underlying signal analysis. CPSP approaches the human signal analysis in a way that is suitable for general signal analysis purposes.

It is our experience that many people see an abundance of additional detail that can be very useful, but that can also be intimidating at first. But generally it is fairly easy to couple system knowledge to the visualization.

All theory on this site is also available in a pdf-file.

[2004] Written by Tjeerd Andringa, Maria Niessen, and Maartje Nillesen