Binaural Sky
The Binaural Sky was designed as a new kind of playback system especially for binaural signals, whereby the techniques of wave field synthesis and dynamic compensation of crosstalk in loudspeaker playback are combined.
The prototype developed in 2005 by Helmut Wittek, Günther Theile and Daniel Menzel at the Institute for Broadcasting Technology (IRT) in Munich consists of a loudspeaker ring fastened over the listener's head. With the help of wave field synthesis , sources are focused close to the ear , whereby ear signals can be reproduced in such a way that a three-dimensional image of sound events is possible without the use of headphones or loudspeakers placed in the field of view.
functionality
Measurement of binaural room impulse responses
The basis of the system is the " binaural room synthesis " (BRS) process, which was also developed at the IRT in the mid-1990s.
A real listening room is measured using a dummy head to give the binaural room impulse response (ger .: binaural room impulse response - BRIR) can detect this very room. In addition to the head-related transfer function (HRTF) of one or more real sources (e.g. loudspeakers) present in the room, this also includes the acoustic properties of the room. With the help of this BRIR, so-called ear signals can be generated, which then convey an impression of the room when played via headphones. This is how a virtual listening room can be synthesized.
Integration of a head tracking system
If the system is implemented in this way, there is a serious disadvantage: Since the determined room impulse response is only valid for a discrete head orientation, the virtual listening room is simply rotated when the head rotates.
In order to prevent this phenomenon, measurements are also carried out in real space with the artificial head rotated. A database is created in which the room impulse responses for a full head turn are stored with a grid of 1 °.
During playback, the current angle of rotation of the head (azimuth) is detected using a "head tracking" system so that the appropriate set of impulse responses from the database can be accessed. With this approach, the virtual listening room remains stable even when the head is turned, and the localization of the virtual sources is also improved.
With the BRS system it is therefore possible to synthesize a familiar listening situation with the help of headphones, or, if the corresponding databases with impulse responses are available, to switch between different listening situations at the push of a button.
Processing of binaural signals for loudspeaker playback
The binaural technology is based on the assumption that the ear signals recorded by the artificial head can be exactly reproduced in the ears of the listener, as this is the only way to create a correct spatial impression. Headphones in particular represent an optimal playback system. If, on the other hand, you want to play ear signals on loudspeakers, you have to consider that the radiated sound fields are superimposed and B. the right ear also receives parts not intended for the left ear, and vice versa.
If at least two loudspeakers are used for playback, the binaural signal can theoretically be filtered on each loudspeaker channel in such a way that the total of the undesired components in the respective ear is canceled out. This procedure is known as crosstalk cancellation (XTC). The biggest problem with the practical implementation is that the XTC filters are only valid for a fixed head position. If the listener only moves his head slightly out of the small listening zone between the loudspeakers or if he turns his head, the sound image is already severely distorted.
This problem could be avoided with an adaptive crosstalk compensation, i. H. the compensation filters are adapted to the current head position, which, however, causes a lot of computational effort. A more elegant solution would be not to change the position of the speakers relative to the head. For this purpose, the sources would have to be able to follow head movements and head rotations synchronously.
Integration of wave field synthesis
This is exactly what wave field synthesis offers: In Binaural Sky, this principle is used to replace the loudspeakers with so-called focused, virtual sound sources. These are generated by a circular loudspeaker array with a diameter of 1 m, which consists of 22 full-range loudspeakers and an additional woofer and is located about 40 cm above the listener's head. The sources can be placed anywhere below this arrangement in the room.
If data are now available about the current orientation of the head, the virtual sources can be kept in a constant position relative to the head or to the ears of the listener with a simultaneous adaptation of the wave field synthesis filter. The entire system can therefore be operated with a single XTC filter, with which a playback method for crosstalk-canceled binaural audio is implemented, which remains stable even when the head is translated and rotated. Figuratively speaking, the result is a “virtual headphone”.
Applications
Basically, the Binaural Sky works like headphones, i. H. There are two output channels, the signals of which are reproduced directly at the right and left ears. But the complete system, i.e. wave field synthesis with crosstalk compensation in combination with binaural room synthesis, is particularly interesting because it opens up fascinating possibilities: Stable three-dimensional monitoring situations can be generated with any source constellation and without disturbing headphones or loudspeakers in the field of view. The listener can thus be placed in a virtual acoustic environment. The system is therefore particularly suitable for "virtual reality" applications or e.g. B. also for computer games.
System performance
The Binaural Sky was presented to the professional world at the Tonmeister Symposium in Hohenkammer 2005 and at the Tonmeister Conference in Leipzig in 2006.
Localization sharpness
A subjective assessment in listening tests showed that the system can deliver a stable spatial image of binaural signals with very good localization sharpness. A problem already known from the BRS is the somewhat less precise localization when the sound events are deflected upwards and downwards (elevation). They are generally rated a little too high by test subjects, images below ear level are associated with larger deviations.
Crosstalk compensation
Measurements show that the system has a good frequency response below about 7 kHz. In addition, aliasing effects of wave field synthesis become noticeable, which primarily reduces the quality of the crosstalk compensation. As a result, the sound is discolored and rated as “perceptible but not disturbing” in listening tests.
Sweet spot
The prototype was initially optimized for a central head position. The system can compensate for head rotations for all angles, but deviations from the center reduce the sound quality as the sources are not tracked and the crosstalk compensation decreases. The listening zone is currently indicated as having a diameter of about 15 cm; if the head is outside of it, playback is automatically muted.
This is also where the prototype's greatest potential for improvement is currently: If the sources can be tracked in real time not only when rotating but also when moving, the listening zone can be greatly expanded.
literature
- D. Menzel: Realization and evaluation of binaural space syntheses using wave field synthesis - Diploma thesis , TU Munich, 2005
- D. Menzel, H. Wittek, G. Theile, H. Fastl: The Binaural Sky: A Virtual Headphone for Binaural Room Synthesis - Tonmeistersymposium 2005 in Hohenkammer
- D. Menzel, H. Wittek, H. Fastl, G. Theile: Binaural space synthesis by means of wave field synthesis - Realization and evaluation - DAGA 2006 in Braunschweig
- P. Kaminski: Binaural Sky - 3D sound over loudspeakers , in: Sound & Recording 3/2006, MM-Musik-Media-Verlag GmbH & Co. KG, Cologne, 2006
- U. Horbach, R. Pellegrini, U. Felderhoff, G. Theile: A virtual surround sound listening room in the OB van - Tonmeistertagung 1998 in Karlsruhe
- U. Horbach, A. Karamustafaoglu, R. Pellegrini, P. Mackensen, G. Theile: Design and Applications of a Data-based Auralization System for Surround Sound - AES convention 1999 in Munich
- T. Lentz, C. Renner: A Four-Channel Dynamic Cross-Talk Cancellation System - DAGA 2004 in Strasbourg
- J. Bauck, DH Cooper: Generalized Transaural Stereo and Applications , Journal of the Audio Engineering Society 44, pp. 683-705,1996
- WG Gardner: 3-D Audio Using Loudspeakers - PhD thesis , Massachusetts Institute of Technology, 1997
Individual evidence
- ↑ U. Horbach, R. Pellegrini, U. Felderhoff, G. Theile: A virtual surround sound listening room in an OB van - Tonmeistertagung 1998 in Karlsruhe
- ↑ U. Horbach, A. Karamustafaoglu, R. Pellegrini, P. Mackensen, G. Theile: Design and Applications of a Data-based Auralization System for Surround Sound - AES convention 1999 in Munich
- ↑ a b c d D. Menzel: Realization and evaluation of binaural space syntheses using wave field synthesis - diploma thesis , TU Munich, 2005
- ↑ a b J. Bauck, DH Cooper: Generalized Transaural Stereo and Applications , Journal of the Audio Engineering Society 44, pp. 683-705,1996
- ↑ D. Lentz, C. Renner: A Four-Channel Dynamic Cross-Talk Cancellation System - DAGA 2004 in Strasbourg
- ^ WG Gardner: 3-D Audio Using Loudspeakers - PhD thesis , Massachusetts Institute of Technology, 1997
- ↑ a b c D. Menzel, H. Wittek, G. Theile, H. Fastl: The Binaural Sky: A Virtual Headphone for Binaural Room Synthesis - Tonmeistersymposium 2005 in Hohenkammer
- ↑ a b c D. Menzel, H. Wittek, H. Fastl, G. Theile: Binaural space synthesis by means of wave field synthesis - Realization and evaluation - DAGA 2006 in Braunschweig