Recent research into the problem of eliminating blur and reverberations from digital signals has led to the invention of a statistical deconvolution method based on the theory of homomorphic filtering. The application described here involves the removal of recording horn resonances from old acoustic discs. The major distortion factors in these discs are surface noise, severely non-flat frequency response produced by the resonances and reverberation characteristics of the recording horn, and non-linear distortion. For a well preserved well recorded copy the non-flat frequency response seems to be the major degrading influence. This degradation shows up in two ways. First, as everyone recognizes, these old recordings are severely lacking in medium to extreme bass and medium to extreme treble. Second, extreme frequency response imbalances are present in the remaining frequency band. Moreover, tests show that for recordings of the human voice the primary subjective degradation is caused by these extreme frequency imbalances which were induced by the old recording horns and the mechanical mechanisms through which the sound waves were carried onto the wax masters. In this restoration effort we have dealt so far with only this last form of distortion.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.