Acquisition of exterior multiple sound sources for train auralization based on beamforming
Regular paper
RWTH Aachen University
Tuesday 2 june, 2015, 17:20 - 17:40
0.1 London (90)
Abstract:
The auralization of train exterior noise requires not only the
identification of
different sound sources, but also the separated signal from each source. A
virtual
sound could not sound convincing if it is created by pure synthesis method
without
actual recordings, especially when there are multiple moving sources.
Beamforming has
been proved qualified to solve the identification and localization problems
of
multiple moving sound sources from trains, however, for auralization these
information
are not enough. This paper focuses on capturing the sound source signals
from a
running train by the method of beamforming for auralization. Beamforming is
applied to
identify the major sound sources and obtain their spatial distribution, and
the
recordings are used as actual signals for the synthesis as well. A
microphone array is
mounted along the railway side to carry out the measurement, which includes
measuring
the exterior noise from different types of trains at different speeds. With
subsequent
post-processing, the main sound sources are localized and identified.
According to the
spectra of the recordings, the sound sources are allocated separately to
tonal and
broadband components. The tones are generated by the propulsion system,
while wheels,
pantograph and coaches contribute mainly to the broadband part, namely
rolling noise
and aerodynamic noise. Finally, it is discussed how these individual signals
can serve
for the auralization of the complete train.