Zoom out Search Issue

IEEE SIGNAL PROCESSING MAGAZINE [29] MARCH 2015
APPLICATION IN ALDs
In [30] and [44], the performance of the binaural MWF and some
of its extensions has been perceptually evaluated, both in terms of
speech intelligibility and localization performance. First, it was
shown that the binaural MWF achieved significant speech intelligi-
bility improvements compared to the bilateral MWF and the bilat-
eral ADM. This demonstrates that transmitting and processing
microphone signals from both devices can result in a significant
gain in noise reduction, especially when multiple interfering
sources are present. Second, using a localization experiment in the
frontal horizontal hemisphere, it was shown that using the bin-
aural MWF with partial noise estima-
tion it is possible to preserve spatial
awareness without significantly
affecting speech intelligibility.
SUMMARY AND OUTLOOK
In this article, we have presented an
overview of several multimicro-
phone signal enhancement algo-
rithms for ALDs and have addressed
other important issues, such as microphone placement and bin-
aural signal presentation. Using appropriate processing with
multiple microphones in a binaural ALD allows both speech
intelligibility improvement as well as a preservation of the audi-
tory perception of the acoustic scene.
Future work in this area will focus both on algorithmic
aspects and a better integration of psychoacoustics. On the algo-
rithmic side, more accurate and robust estimation and careful
exploitation of comprehensive spatiotemporal signal statistics
for all relevant sources in highly time-varying scenarios will be
necessary to allow for the ultimate desired binaural presenta-
tion. The learning of acoustic scenarios and source characteris-
tics can certainly be expected to contribute to reaching this
goal. Optimum distribution of the computational load over the
available computing hardware via bit rate-constrained “body
area networks” will constitute another challenge to algorithm
developers. On the psychoacoustic side, ideally, meaningful cri-
teria are desirable that can directly be integrated into the cost
functions to allow perceptually optimum signal processing at
any given time instant. This may start from incorporating gen-
eral knowledge about well-known noise masking effects com-
bined with knowledge on the relative importance of certain
binaural cues as used already in audio coding and reach to more
powerful, yet unknown models for human hearing. For each
individual, it should be merged with knowledge about possible
hearing impairments or personal listening preferences, i.e., a
so-called auditory consumer profile. One may speculate that
with suitable user interfaces, the traditional fitting procedures
will be replaced by training procedures supervised by the user
and even the cost functions for optimizing the multichannel fil-
tering will be as individual as the users themselves. All of these
developments will certainly benefit from the integration into
handy, but powerful personal computing platforms that are
already emerging.
ACKNOWLEDGMENTS
This research was supported in part by the Research Unit FOR
1732 “Individualized Hearing Acoustics” and the Cluster of Excel-
lence 1077 “Hearing4All,” funded by the German Research Foun-
dation (DFG), and by the European Union under contract
FP6004171 (HEARCOM).
AUTHORS
Simon Doclo ([e-mail]) received the Ph.D.
degree in applied sciences from Katholieke Universiteit Leuven,
Belgium, in 2003. Since 2009, he has been a full professor at the
University of Oldenburg, Germany,
and a scientific advisor for the Fraun-
hofer Institute for Digital Media
Technology. His research activities
center on signal processing for
acoustical and biomedical applica-
tions. He received the EURASIP Sig-
nal Processing Best Paper Award in
2003 and the IEEE Signal Processing
Society (SPS) Best Paper Award in
2008. He was member of the IEEE SPS Technical Committee on
Audio and Acoustic Signal Processing and is an associate editor of
the EURASIP Journal on Advances in Signal Processing.
Walter Kellermann ([e-mail]) received the Dipl.-Ing.
degree in electrical engineering in 1983 and the Dr.-Ing. degree in
1988. From 1989 to 1990, he was a postdoctoral member of tech-
nical staff at AT&T Bell Laboratories. He has been a professor of
communications at the University of Erlangen-Nuremberg,
Germany, since 1999. He (co)authored 16 book chapters and more
than 200 refereed papers and was corecipient of several best paper
awards for IEEE publications. He served as an associate editor and
as a guest editor to various journals and was a general chair for
several international conferences. His service to the IEEE Signal
Processing Society (SPS) includes Distinguished Lecturer, chair of
the IEEE SPS Technical Committee on Audio and Acoustic Signal
Processing, and member-at-large for the SPS Board of Governors.
He is an IEEE Fellow.
Shoji Makino ([e-mail]) received the Ph.D.
degree from Tohoku University, Japan, in 1993. Since 2009, he has
been a professor at the University of Tsukuba. His research inter-
ests include acoustic signal processing for speech and audio appli-
cations. He received the ICA Unsupervised Learning Pioneer
Award and the IEEE Machine Learning for Signal Processing
Competition Award. He has served on the IEEE Signal Processing
Society (SPS) Technical Directions Board, Awards Board, and
Conference Board. He was associate editor of IEEE Transactions
on Speech and Audio Processing. He is the chair of the IEEE SPS
Technical Committee on Audio and Acoustic Signal Processing.
He is an IEEE Distinguished Lecturer and an IEEE Fellow.
Sven Nordholm ([e-mail]) received the Ph.D.
degree in signal processing from Lund University, Sweden, in 1992.
Since 1999, he has been a full professor at Curtin University, Perth,
Australia, and an advisor for Hearmore and Sensear. His research
activities focus on signal processing and communication in
ALL OF THESE DEVELOPMENTS
WILL CERTAINLY BENEFIT FROM
THE INTEGRATION INTO HANDY,
BUT POWERFUL PERSONAL
COMPUTING PLATFORMS THAT
ARE ALREADY EMERGING.
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
q
q
M
M
q
q
M
M
q
M
THE WORLD’S NEWSSTAND
®
Previous Page | Contents | Zoom in | Zoom out | Front Cover | Search Issue | Next Page
q
q
M
M
q
q
M
M
q
M
THE WORLD’S NEWSSTAND
®
__________________
_______
_______________
________________