Conference PaperPDF Available

A Mobile App for Room Acoustical Measurements

Authors:
A Mobile App for Room Acoustical Measurements
Andreas Rosenkranz, Ralf Burgmayer, David Ackermann, Markus H¨adrich und Stefan Weinzierl
TU Berlin, FG Audiokommunikation, stefan.weinzierl@tu-berlin.de
Introduction
The assessment of room acoustical parameters is tra-
ditionally conducted with PC-based software or specif-
ically designed measurement devices. Recently, however,
mobile devices offer ever-increasing processing power, a
graphical user interface and network connections. Thus,
they seem to be perfectly suited for the determination of
room acoustical properties. Since these smart devices are
also commonplace, turning them into measurement tools
could make additional tools unnecessary for quick and
spontaneous room acoustical assessments. This contri-
bution presents a multi-platform, easy to use, ISO com-
pliant open-source measurement system to assess room
acoustical properties and quality of speech transmission
in a room. The measurement system is implemented as
an installable hybrid mobile application.
Room Acoustic Parameters
The determination of room acoustic parameters is based
on room impulse responses recorded at 44100 Hz sam-
pling rate, which can be reduced to 22050 Hz to address
the processing capabilities of older mobile devices. The
recording is automatically triggered by an input signal
exceeding a given threshold above a noise level estima-
tion made at the start of a new measurement. The room
impulse response is then recorded and stored for later ex-
port as a 16 bit PCM WAV file containing all measured
impulse responses for a single measurement position. All
results calculated by the application can also be exported
as a preformatted text file, that can be easily imported
into spread sheet applications such as Microsoft Excel or
Apache Open Office Calc.
The room acoustic parameters included in the develop-
ment of the app so far are reverberation time (RT20,
RT30), bass ratio, clarity (C50, C80), definition (D50,
D80), center time [1] and the speech transmission index
(STImale, STIfemale ) [3]. Their calculation is based on a
room excitation by impulsive stimuli and is supported
by a number of features such as an automated start- and
endpoint truncation of room impulse responses as well as
energy compensation for the truncation and an iterative
SNR assessment [4]. An octave and third-octave band
IIR filter bank [2] for signal analysis was implemented,
as well as the option to use time reverse and zero phase
filtering [5] for narrow filter bandwidths and short rever-
beration time measurements.
Supported Platforms
The application is available as an installable mobile app
as well as as a web application for offline local use or re-
mote online use, when hosted on a web server. It was de-
veloped as a cross-platform approach and thus supports
the mobile platforms iOS starting with iOS 9 and An-
droid starting with Android 5. A browser-based instance
of the application is also available for Mozilla Firefox
starting with v50 as well as Google Chrome v49+.
Comparison to Existing Systems
To evaluate the suitability of the application as a
measurement tool, it was deployed on a smart device
(Apple iPhone 4S). A variety of measurements were
made and compared to the results of different room
acoustic toolboxes as well as a measurement device
manufactured by NTi Audio. All measurements on
the smart device were conducted with the MicW i436
measurement microphone specifically developed for the
use with mobile phones and tablets [9]. It is certfied as
a class 2 calibrated measurement microphone according
to IEC 61672.
For the first measurement a class room of the TU
Berlin (MA005) was excited with a starter clap. Five
impulse responses were recorded at the same position
and truncated at the starting point 20 dB below their
peak. The truncated recordings were averaged into a
mean impulse response to obtain a higher signal-to-noise
ratio and reduce the computational costs. This mean
impulse response was filtered by a twelfth order octave
band filter in bands ranging from 125 Hz to 8 kHz. The
reverberation times extracted from the band limited
impulse responses were compared with values from the
ITA Toolbox [8] and the AARAE Toolbox [7], and
delivered almost identical results (Figure 1).
Figure 1: RT20 values calculated from measured room im-
pulse responses with the developed mobile app, with the ITA
toolbox and with the AARAE Toolbox. Excitation was done
with a starter clap
DAGA 2017 Kiel
235
A second measurement was conducted in a living room.
The excitation was done via hand clap. Five room
impulse responses were recorded and again truncated
20 dB below their peak. A mean impulse response
was calculated and filtered in octaves. An endpoint
truncation according to the algorithm proposed by
Lundeby et al. [4] was done without truncation energy
compensation. Subsequently the reverberation time
RT20 was calculated by the presented applications
algorithms and the two aforementioned toolboxes. The
results can again show only minor deviations (Figure 2).
Figure 2: RT20 values calculated from measured room im-
pulse responses with the developed mobile app, with the ITA
toolbox and with the AARAE Toolbox. Excitation was done
with a hand clap.
External versus internal microphone
Figure 3 shows the results of a reverberation time
measurement in a living room. It was conducted with
the MicW i436 as well as with the internal microphone of
the smartphone in the same position. In the 125 Hz and
250 Hz octave bands the signal-to-noise ratio delivered
by the built-in electret microphone with its non-linear
frequency response [6] was obviously not high enough
to obtain reliable RT20 values. The small differences
at higher frequency bands are likely to be due to the
directional characteristics of the internal microphone,
as opposed to the omni characteristic of the external
measurement microphone.
Mobile app versus NTi XL2 system
Since the mobile application was built for easy use and
fast measurements, a comparison measurement with the
NTi XL2 system was conducted [10] (Figure 4). The
NTi XL2 was used in combination with the NTi 2230
measurement microphone. The smart device was again
used with the MicW i436 external microphone. The
measurement was conducted in a class room that was
excited with a starter clap. Simultaneous recordings
were conducted with the microphones of both systems
positioned as close as possible to each other. The NTi
XL2 system did not provide results below 500 Hz due to
Figure 3: RT20 values calculated from measured room im-
pulse responses obtained with the internal microphone of the
smart device and the MicW i436 external class 2 measurement
microphone.
the low energy in these bands given by the form of exci-
tation, so they were left out of the comparison. As shown
in Figure 4, the measured reverberation times differ by
a maximum of 60 ms between the two measurement
systems. It is unclear whether the differences are due
to slightly different microphone characteristics, different
algorithms for the calculation of the reverberation time,
the different sampling rate (the NTi XL2 operates at
48000 Hz) or the slightly different recording positions.
Figure 4: RT20 values obtained with the mobile application
and the NTi XL2 system.
Discussion
In the development of the application, the limited capa-
bilities, notably of older smart devices, have proven to
be a source of potential problems, since the number of
impulse responses to be kept in-memory during the cal-
culation process was limited by their internal memory.
This is particularly problematic for the filtering process
in third-octaves and the subsequent truncation of the fil-
tered signals, but has been addressed by optimising the
DAGA 2017 Kiel
236
number of impulse responses residing in memory at the
same time and can further be improved by reducing the
sampling rate to 22050 Hz.
The effects of the use of internal microphones across a
wide variety of smart devices and their directional charac-
teristics have yet to be studied systematically. However,
already with the smartphone used in the test, the high-
pass characteristics of the internal microphone caused
problems with the excitation of low-frequency reverber-
ation times. Thus, the use of external measurement mi-
crophones is recommended.
Targeting multiple platforms with a single code base in
the cross-platform development is slower and requires
more processing power than a native approach on each
single platform. Nevertheless, the convenience of a single
code base for an open-source project will facilitate the
further development of the mobile application, since ev-
ery contribution to the project only has to be made only
once and can then be deployed to all platforms.
The open-source approach provides a convenient plat-
form to add new features as well as customizing already
implemented features such as filters, onset detection and
general audio processing. The first tests shown above
suggest the presented application to be a suitable mea-
surement tool. It provides the possibility of quickly as-
sessing room acoustical properties with complete ISO
compliance, when combined with a certified measurement
microphone. With its open-source and cross-platform ar-
chitecture, it can be particularly attractive for the use in
an academic environment.
References
[1] DIN EN ISO 3382: Acoustics - Measurement Of
Room Acoustic Parameters. Beuth Verlag, Berlin,
October 2009.
[2] DIN EN 62160-1: Electroacoustics - Octave-Band
And Fractional Octave-Band Filters. Beuth Verlag,
Berlin, October 2014.
[3] DIN EN 60268-16: Elektroakustische Ger¨ate - Teil 16:
Objective Rating Of Speech Intelligibility By Speech
Transmission Index. Beuth Verlag, Berlin, May 2012.
[4] Lundeby, A., Vigran, T. E., Bietz, H., and Vorl¨ander,
M.: Uncertainties of measurements in room acoustics.
Acta Acustica united with Acustica 81 (1995), 344-
355.
[5] Jacobsen, F., and J. H. Rindel: Time Reversed De-
cay Measurements, Journal of Sound Vibration 117
(1987), 187-190.
[6] Fischer, J., Mayer, T., Geng, N.: Evaluation of
Smartphone-based Audio Applications. Fortschritte
der Akustik, AIA-DAGA 2013 Merano, 1701-1702..
[7] Carbrera, D., Jimenez, D., Martens, W.L.: Au-
dio and Acoustical Response Analysis Environment
(AARAE): A Tool To Support Education and Re-
search in Acoustics, Proceedings of the Internoise,
2014.
[8] Dietrich et al.: ITA-Toolbox – An Open Source
Matlab Toolbox for Acousticians, Fortschritte der
Akustik, DAGA Darmstadt 2012, 151-152.
[9] Mic W R
Professional Microphones: http://www.
mic-w.com
[10] XL2 Sound Level Meter & Acoustic Analyzer - NTi
Audio
DAGA 2017 Kiel
237
ResearchGate has not been able to resolve any citations for this publication.
Article
The influence of several sources of error on room acoustical measurements have been investigated using maximum-length sequences (MLS). The algorithms for the determination of room acoustical parameters used by different analyzers introduce systematic differences caused by differences in time windowing and filtering, in reverse-time integration and in noise compensation. The overall uncertainty of measurements is of the same magnitude or a little higher than subjectively perceivable changes in room acoustical parameters when the measurements are performed according to ISO/DIS 3382. However, the draft standard allows various procedures to be applied in the processing of impulse responses.
Evaluation of Smartphone-based Audio Applications. Fortschritte der Akustik, AIA-DAGA 2013 Merano
  • J Fischer
  • T Mayer
  • N Geng
Fischer, J., Mayer, T., Geng, N.: Evaluation of Smartphone-based Audio Applications. Fortschritte der Akustik, AIA-DAGA 2013 Merano, 1701-1702..
Audio and Acoustical Response Analysis Environment (AARAE): A Tool To Support Education and Research in Acoustics
  • D Carbrera
  • D Jimenez
  • W L Martens
Carbrera, D., Jimenez, D., Martens, W.L.: Audio and Acoustical Response Analysis Environment (AARAE): A Tool To Support Education and Research in Acoustics, Proceedings of the Internoise, 2014.