Last month, I went to Oldenburg to conduct an oral history interview with Kunstkopf-pioneer Volker Mellert (read more about my interview and the Oldenburg-dummy head here). As professor emeritus for applied physics (acoustics) at Carl von Ossietzky University, Mellert is still involved in several research projects. One of these projects is located at Jade Hochschule (another Oldenburg based university): it aims at developing a virtual dummy head. As part of my visit in Oldenburg, I got the chance to listen to this virtual dummy head.

The idea behind it is that for ideal binaural recordings one would need (at least theoretically) individual dummy heads as we all have different head sizes and forms and diverse ear shapes. Researchers have long tried to construct something like the best “average” dummy head but for perfect binaural transmission we still need to listen to/with our own ears. The new approach wants to provide individual dummy heads – but virtual ones. The virtual dummy head, built by Matthias Blau and his team, is based on a three step process: first, the individual head-related transfer function (hrtf) of a test person is measured (and they are currently developing a method to simplify the necessary measuring procedure); secondly a special microphone array is used to record a particular sound event; and thirdly, an individualized binaural version of this recording is calculated based on the test person’s personal hrtf.

Below you can see the microphone array in a sound studio at Jade Hochschule. The array consists of 24 microphones, and they use a Golomb ruler to make sure that no two microphones are the same distance apart (here you can find a detailed description of the microphone positioning). The array shown in my photos is optimized for spatial recordings in the horizontal plane; however, the next step will be to build an array for spatial recordings of the upper half-space.

To experience the virtual dummy head, the researchers have built a simple software front-end: here you can select between four different hrtf (individual hrtf of four members of the research team) or bypass the virtual dummy head and listen to the “original” recording. Equipped with a pair of headphones I first listened to a live demonstration of the virtal dummy head and then to a virtual dummy head recording of an organ concert in an Oldenburg church. During the live demonstration one of the researchers walked around the microphone array in the studio while I sat in the control room and switched between the four hrtf to experience the different spatial impressions. During this spatial hearing test, and even more when I listened to the organ concerto, I was most amazed by the audible differences between the four hrtf. Three of them gave me fairly good spatial impressions and during the live I could well discriminate the sound directions, but one hrtf sounded to my ears muffled as if I had my hands on my ears; obviously this hrtf was very different from my personal one. From this experience one can better imagine that conventional dummy head recordings sound very bad to some ears and give only poor spatial impressions. It would be intriguing to listen to the virtual dummy head with your own hrtf as base for the binaural sound calculation. At the moment, the researchers aim at developing a professional system for industrial sound design; but another idea is to build a system that would allow you to use your personal hrtf for binaural sound experience in 3D gaming.

Advertisements