Among the many structures in pinna, it is well known that
the posterior wall of the concha is responsible for the spectral
notches in HRTFs that serve as important front–back and up–down
cues [16,17,23,26]. Hebrank and Wright [16] hypothesized that
reflections from the posterior wall of the concha alone may be
responsible for the observed notch in the median plane, and
they proposed a simple reflection model to derive the spectral
notches. More recently, Lopez-Poveda and Meddis [24] proposed
a diffraction/reflection model of the concha wall to derive the
notch frequencies for elevated sources on vertical planes, and they
reproduced the spectral notches more accurately. Inspired by the
previous studies, we describe the two kinds of artificial robot
ear design mimicking the human concha. Bilateral asymmetric
microphone position is also investigated to enhance the estimation
performance.