|
Lead-in
Our ear is a complex machine and involves many different processes. They
enable us to hear specially and to pinpoint a sound source in the space
around us. In cooperation with the institute for broadcast engineering, IRT
Munich, the Phoenix 5.1 was developed to acoustically represent a virtual 5.1
studio setup in a headphone by simulating a multichannel monitor room in a
sub optimal environment (e.g. broadcast van, home cinema in living room), all
in professional quality.
The primary target application of the Phoenix 5.1 is mainly in professional audio.
A sound engineer in a  broadcast van, for example, is able to do a “live”-
mixdown at a place, which is not really suitable for a mixdown because of its
acoustical properties. The sound engineer is mixing like in a normal multichannel
surround studio. Instead of the 5+1 speaker system, which is the sound source
in a normal studio mixdown, the sound engineer works with the Phoenix system only
through headphones, which generate the virtual environment in connection with
the BRS technique. An important feature of the the Phoenix 5.1 system is its
ability to accurately detect any head movements and to continuously adjust
the audio processing so as to realistically simulate a stationary
multichannel sound source in a space where the listener can move his / her
head. From a technical point of view the audio signals are convoluted with
measured (real) room impulse responses.
|
|
Listening
with headphones
Compared with monitor speakers,
headphones offer with much greater efficiency at least the same sonic
qualities, such as bandwidth, freedom of linear and non-linear distortions
and also the same maximum sound pressure at the ear. Furthermore, the
acoustical behavior of the listening room has no influence on the playback performance
through headphones. However, without any manipulation, the well known “head
internal localization” of the listening event can be a disadvantage.
Operating
mode of BRS
 For the BRS process, an existing monitor room with its
speakers is accurately measured with a dummy head positioned in the reference
listening position. The processor generates the desired signals, such as the
left speaker for instance, through convolution of the source signal with the
appropriate HRTF pair. The calculated output signals for the headphones are
similar to the signals as measured on the dummy head in the sweet spot of the
studio. Provided the parameters of the dummy head and the headphones are well
defined and considered in the process, the BRS processor provides the same
listening experience as perceived by the dummy head. To prevent any outer ear
and head dependend front/rear inversions, spontaneous head movements are also
considered in the process. For this, a head-tracking system transmits the
head position signals to the BRS processor which then allocates dynamically
the correct HRTF (outer ear transfer function) that is stored in the BRS
data. With the BRS pro cessor it is now possible to simulate a realistic
monitoring situation in multichannel.
|
|
Diffuse-field
equalization
The headphones are an integral
part of the whole system. The headphones get diffuse-field equalized, just
like the dummy head too. As the natural binaural function of the outer ear
becomes ineffective when wearing headphones, they have to replace that
function in the diffuse-field. By equalizing the summation of all direction
specific linear distortions, the influence of the outer ear on the sonic
quality gets minimized as well.
Latency
Time
 The latency is defined as the time between impulse
and response, in this case head rotation and data processing. If the message
of a fast head movement to the right gets delayed too much on its way to the
BRS, then the listener perceives this as a displacement of a stationary sound
source to the right. As soon as the head movement then gets processed, the
perceived location of the sound moves back again to its original location.
Therefore, the latency needs to be kept as short as possible, which means that
more head movements need to be processed in shorter periods of time. As this
increases processing needs, an optimal balance had to be found between sound
performance and efficient processing power. The resulting architecture is
able to avoid any audible artefacts that can result from rapid changes. To
speed up the processign, corresponding spectra are stored in fast RAM.
Depending on the head position the necessary spectrum is calculated from an
efficient working algorithm.
|