Implementation eines Phasenvocoders zur Klangmodifikation

The main purpose of this work is the simulation of masking phenomena of the human auditory system by a software and the improvement of existing models. Two models for sound modifications were implemented on a NeXT-computer by using the phase vocoder as analysis/synthesis technique. With this technique an explicit separation of temporal and spectral information is possible. Thus, a sound can be modified by time scaling and (or) pitch transposition and (or) transposition of the formants. The parameters for the analysis and the type of the modification with ist parameters are variable. The sound is representable in a spectogram and can be modified in the time-frequency domain.
With the second model it is possible to simulate the simultaneous and forward masking properties of the human ear. To determine an optimal approximation of these masking phenomena by experiments there is a large amount of chooseable parameters. A threshold is produced to separate the relevant form the irrelevant signal components. The latter can be added to or removed from the signal without changes in the auditive perception. Components determined as irrelevant by the masking phenomenon, are eliminated from the signal frame.