Investigation of psychoacoustic principels for automatic mixdown algorithms

Diploma Thesis (pdf. 4.631 KB)

n the course of the recording process, the sound field of one or more instruments is transformed from the mechanic into the electric domain by one or more electroacoustic transducer. If you electronically sum these signals to any kind of mixdown format (e.g. Stereo or 5.1), on the one hand the aim is to picture natural sound fields as close to reality as possible, on the other hand unreal acoustic scenarios are created consciously or unconsciously. The paper at hand develops an evaluating description for different scenarios, relating to tested perception psychological effects.

It is for instance possible to look at the spectrum of a signal and deduce its masking characteristics, i. e. calculate the hearing thresholds that variable concerning time and frequency. If you mix a signal to a second signal, parts of this newly added signals will be audible, while others will remain inaudible. But also the newly added signal works as a masker and covers or reduces parts of the first signal. These proportions are dependent on enormous fluctuation in time and frequency. That is why it has to be graphically and quantitatively visualized by means of signal theoretical and statistical tools. Also the positioning in the panorama of the mixdown has effects on these masking relations and can be taken advantage of during the mixdown.

It is the aim to win reference variables for a control algorithm that controls processors for influencing the mixdown. Such processors can be used for carrying out an automatic and adapted frequency separation of the different musical instruments. In this process, technical and aesthetic quality criteria should equally be fulfilled and an effective human-machine interface should be created.