IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 60, NO. 4, APRIL 2011 1141
Instantaneous Binaural Target PSD Estimation
for Hearing Aid Noise Reduction in
Complex Acoustic Environments
A. Homayoun Kamkar-Parsi and Martin Bouchard
Abstract—Some hearing aid models such as the in-the-ear or
in-the-canal model are small, and due to size constraints, only
a single microphone per hearing aid can be fitted. As a result,
only the single-channel noise reduction schemes can be integrated
in them. However, in the near future, binaural hearing aids will
be available. These hearing aids will allow hearing devices on
each side of the head to receive information from the micro-
phones located on both devices via a wireless link. This addi-
tional information allows the use of joint statistics and/or spatial
information, which allows the use of better denoising methods.
The main contribution of this paper is to introduce a binaural
target power spectral density (PSD) estimator for binaural hearing
aids under the presence of a background interfering talker (or
a lateral transient noise). This paper then presents an example
of how the proposed target PSD estimator and another recently
developed binaural diffuse noise PSD estimator can be integrated
to produce a binaural noise reduction scheme that can operate
in real-life complex acoustic environments that are composed of
time-varying diffuse noise, multiple directional nonstationary
talkers (or noises), and reverberant conditions. Such a binaural
noise reduction system allows the substantial reduction of different
combinations of diverse background noises and the increase in
speech intelligibility while guaranteeing the preservation of the
interaural cues of both the target speech and the directional back-
ground noises. A comparison with several advanced denoising al-
gorithms is made in this paper by using several objective measures
and real-life hearing aid signals, showing the good performance
of the example binaural noise reduction scheme that uses the
proposed PSD estimator.
Index Terms—Binaural estimators, binaural hearing aids,
interaural cues preservation, nonstationary noise reduction,
speech enhancement, target power spectrum density (PSD)
estimation.
I. I NTRODUCTION
F
OR HEARING aid devices with two or more microphones,
the microphone array technology can provide great ben-
efits. The microphones can be configured in a small endfire
Manuscript received December 16, 2009; revised May 27, 2010; accepted
August 25, 2010. Date of publication November 1, 2010; date of current version
March 8, 2011. This work was supported in part by a Natural Sciences and
Engineering Research Council of Canada (NSERC) student scholarship and in
part by an NSERC-CRD research grant. The Associate Editor coordinating the
review process for this paper was Dr. Jiong Tang.
A. H. Kamkar-Parsi is with the Siemens Audiologische Technik GmbH,
91058 Erlangen, Germany (e-mail: hkamkarparsi@gmail.com).
M. Bouchard is with the School of Information Technology and Engineering,
University of Ottawa, Ottawa, ON K1N 6N5, Canada (e-mail: bouchard@
site.uottawa.ca).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIM.2010.2084690
array, which allows the implementation of classical beamform-
ing schemes. Speech enhancement or denoising using beam-
forming takes advantage of the spatial diversity of the target
speech and noise sources by altering and combining multiple
noisy input microphone signals in a way that can significantly
reduce background noise and increase speech intelligibility.
Unfortunately, due to size constraints, only certain hearing
device models such as behind-the-ear (BTE) can accommodate
two or occasionally three microphones. Smaller models such
as the in-the-canal (ITC) or in-the-ear model only permit the
fitting of a single microphone. Consequently, beamforming
cannot be used for such cases, and so far, only the single-
channel noise reduction schemes could be used (i.e., with a
single microphone per hearing device). These single-channel
noise reduction schemes are less effective since the spatial
information cannot be exploited, and they also tend to introduce
some speech distortion or musical noise, especially if the noise
reduction is aggressive.
In the near future, new types of high-end hearing aids such as
binaural hearing aids will be available. In the current bilateral
hearing aids, a hearing-impaired person wears a hearing aid on
each ear, and each hearing aid processes only its own micro-
phone input. Unlike these current systems, the new binaural
hearing aids will allow hearing devices on each side of the head
to receive information from the microphone located on both
devices via a wireless link. As a result, working with a binaural
system, new classes of noise reduction schemes as well as new
power spectrum density (PSD) estimation techniques can be
developed [1]. The previous few attempts to include binaural
processing in hearing aid noise reduction algorithms have not
fully achieved a significant improvement in performance under
difficult practical situations. In [2] (which complements the
work in [3] and several related publications such as [4] and
[5]), a binaural multichannel Wiener filter (MWF) technique
with a modified cost function was developed to essentially
reduce directional noise while having some control over the
distortion level of the binaural interaural cues for the target
speech and noise components. A good performance for the
MWF was reported for the case of a single directional noise
in the background [2]. However, the directional background
noise is restrained to be stationary or slowly fluctuating for
the MWF to perform well, and the noise source should not
relocate during target speech activity since its characteristics
are only computed during target speech pauses. In addition, for
cases where the directional noise is speech or when there are
0018-9456/$26.00 © 2010 IEEE