Distortion of Musical Signals by Means of Multiband Waveshaping
Pablo Fern�ndez-Cid, Javier Casaj�s Quir�s · Journal of New Music Research · 2001
In this work we propose a new architecture for musical distortion. WaveShaping (a sound synthesis technique from the mid-late 1970s) is used here as the basic distortion generation element. WaveShaping can easily and precisely generate upper partials from a flat-envelope pure sinusoidal signal. The extent and distribution of the overtones series can be adjusted at will. In order to apply WS techniques to distort real world audio signals, a multiband front-end is used. The front-end filterbank extracts simple (ideally monotonal) nonfull band signals, whose amplitude envelope is flattened previous to WS. This way, WS can be adjusted (to set the relative magnitude of newly generated overtones and to avoid aliasing components) and applied in a per-band basis. As another benefit, the structure makes intermodulation distortion less of a problem, balancing the end result towards “harmonic” distortion rather than “metallic” or “buzzing”. The proposal must be understood in terms of a “distortion- synthesiser” rather than in terms of detailed emulation of old equipment or physical models of non-linear devices, and allows the musician to adjust many parameters in a musically meaningful domain. Rich, detailed and highly personal sounding distortions can be created by the end user and can be imposed over real world signals.