Skip to content

Aktuelles Kunst und Forschung Lehre Mediathek Services & Info IEM - intern Contact
  Sie sind nicht eingeloggt. Link icon Log in
Sie sind hier: Startseite » Kunst & Forschung » Akustik » Psychoakustik » Aufbereitung von kopfbezogenen Übertragungsfunktionen für virtuelle binaurale Mehrkanalsysteme

Aufbereitung von kopfbezogenen Übertragungsfunktionen für virtuelle binaurale Mehrkanalsysteme

Harald Ziegelwanger

When presenting sound sources via earphones, head-related transfer functions (HRTFs) allow to position the sounds in the virtual space around the listener. HRTFs describe the filtering of the incoming sound due to the pinna, head, and torso, and include both broadband delays and spectral cues. The individually-measured HRTFs can provide realistic spatialization of sounds in terms of a high vertical-localization accuracy and good externalization of the sounds.

Measured HRTFs are limited at lower frequencies mostly due to the band limitations of the loudspeakers. However, the lower frequencies, down to 50 Hz, seem to be necessary for naturally-sounding sounds. Thus the band-limitation has to be compensated by augmenting the HRTFs after the measurements. While the augmentation of the spectral amplitudes appears to be trivial, the phase augmentation requires more attention. Because of the direct link between the phase and the interaural time differences (ITDs), the main cue for sound localization in the horizontal plane, inaccuracies in the phase compensation may affect the perceived horizontal sound position. Hence, the augmentation must be applied considering the sensitivity to ITDs in the lower frequencies.

For an efficient spatialization of virtual sound sources, HRTFs are usually separated into broadband group delay and minimum-phase systems. While previous studies were focused on ITD reconstruction, for the reproduction of multichannel systems, the group delay must be estimated independently for each ear to preserve the delay association between the multiple spatial sources.

Room reverberation also affects the quality of spatialization. In its simplest form, the reverberation can be simulated by using mirror sources with the order and position accuracy of the mirror source as the most salient parameters of the model.

The main goal of this thesis is to develop an automated algorithm to pre-process individually-measured HRTFs for multichannel binaural reproduction systems with room simulation. In this algorithm, HRTFs will augmented and encoded by extracting broadband delay. The effect of the involved parameters will investigated in psychoacoustic experiments. The algorithm will be implemented and incorporated into the software framework ExpSuite.

Also, a real-time technology-demo will be implemented to provide results as open-source software, with a simple room-simulation to enhance the quality of the final spatialization.

Harald Ziegelwanger    Typ: Diplomarbeit    Status: Projekt läuft     Datum: 01.03.2011

Zuletzt verändert: 30.03.2011