TU Berlin

Audio Communication GroupDynamic binaural synthesis

Logo Fachgebiet Audiokommunikation

Page Content

to Navigation

Dynamic Binaural Synthesis

Description of the project

Dynamic Binaural Synthesis

Binaural snythesis allows a realistic rendering of three-dimensional sound fields through headphones. Comparable with the visual impression that is produced in 3D cinema, sensations of sound sources at arbitrary positions in real or artificial spaces can be generated. This means in particular that sound sources such as musical instruments or speaker can be perceived outside the head of a listener, although for playback of binaural signals only headphones are required.

For this purpose so-called binaural room impulse responses - which can be thought of as the acoustic fingerprints of sound sources in space - are either measured using an artificial head which is equipped with microphones instead of the eardrums or simulated using dedictated room acoustic software. These impulse responses can then be used to filter any acoustic signal. When listening through headphones the auditory impression of the sound source reproducing the selected signal in this room is evoked.

For research purposes, but also for room acoustics design, it is particularly interesting that different rooms or different design conditions with virtually any acoustic signals can simultaneously be assessed.

A particularly high degree of realism of the simulation is achieved in that it responds properly to the head movements of the listener. For this purpose the position of the head of the listener is determined continuously and in real time. The head position data is analyzed by the simulation software, and the filters used - the binaural impulse responses - are adapted in an inaudible fast manner.
Therefore, a listener can turn his head to the left while a sound source - e.g. originally percieved from the front and outside of the head - seems to maintain its correct position, rather than - as is known from conventional headphone reproduction - co-rotate with the listeners head. In this case, one speaks of dynamic binaural synthesis [1].

The Binaural Reference Simulation System of the Audio Communication Group

For conducting listening tests and for demonstration purposes, a comprehensive framework for dynamic binaural synthesis has been developed by the Audio Communication Group:

The listener controls the binaural simulation via a network capable graphical user interface [2] which enables instantaneous switching between simulations of e.g. different rooms. If a listening test is to be conducted, the interface is  can be used to automatically save the test subjects' responses.

For simple switching between different simulations, anechoic audio material [3] is filtered with different pre-produced data sets of binaural impulse responses [4]. This is done by a mathematical calculation rule - the so called fast and time variant convolution. The application "fWonder", which was developed by the Audio Communication Group, is used for that purpose [1].

In addition to head movments of the listeners, the binaural simulation also takes into account the their head sizes, because - due to individual ear distances and resulting so-called interaural time difference - the latter significantly influences the perceived location of a sound source [5]By a continious individual adjustment it is ensured that the localization of a sound source - especially during head movements of the listener - not changed erroneously.

For achieving a realistic simulation it is further important to remove distorting influences of the headphones. The specially developed 'P.R.E.C.I.S.E.' method is applied to individually or generically compensate headphone transfer functions which have been measured with insert microphones [6].

For optimum reproduction, the BK211 headphones system is used, which was developed specifically for binaural reproduction by the audio communication group [7]. The associated amplifier produces an extremely low self-noise. The inculded digital signal processors deliver a basic acoustic correction , matching crossover filters for combined playback with a subwoofer, and - if desired - a diffuse-field equalization suitable for conventional high-fidelity headphone reproduction.


Quick Access

Schnellnavigation zur Seite über Nummerneingabe