Abstract (eng)
This thesis presents a comparison of deconvolution-methods for impulse responses of loudspeakers. To explore the topic, the processing of the human ear in the time domain is discussed, to derive the importance of time patterns for sound reproduction. Three basic methods of deconvolution are discussed. Aside from those the state of research with variations of these methods is discussed. The discussion of the state of research includes experiments to determine the processing of the human ear in the time domain and hence the significance of the impulse response. Results are not conclusive, some researchers favor the amplitude frequency response, others the impulse response.
To enable a loudspeaker to have an approximation of a dirac as impulse response, it is imperative for the audio crossover to have almost an dirac as impulse response; i. e. the sum of low-, band- and highpass has to be an impulse. The author ist designing a correseponding audio crossover with bessel-filters. To test the different deconvolution methods, the author creates scripts in scilab and measures impulse responses of different tweeters.