Professional Documents
Culture Documents
This paper was peer-reviewed as a complete manuscript for presentation at this Convention. This paper is available in the AES E-
Library, http://www.aes.org/e-lib. All rights reserved. Reproduction of this paper, or any portion thereof, is not permitted without
direct permission from the Journal of the Audio Engineering Society.
ABSTRACT
This research addresses the usefulness of an absolute descriptor to quantify the degree of diffusion in a third
octave band basis of a sound field. The degree of sound field diffuseness in one point is related with the
reflection’s energy control multiplied by the temporal distribution uniformity of reflections. All this
information is extracted from a monaural, broadband, omnidirectional, high S/N impulse response. The
coefficient range varies between 0 and 1, evaluates the early, late and total sound field, for frequencies above
Schroeder’s and in the far field from diffusive surfaces, zero being “no diffuseness” at all. This coefficient
allows the comparison of different rooms, different places inside rooms, measurement of the effects of different
sound diffusers coatings and the resulting spatial uniformity variation, among other applications.
Bidondo et al. Measuring Sound Field Diffusion: SFDC
is the average of the SFDC by bands results. It describes Certainly, the SFDC is not related to the length of the
the average reflection’s amplitude control and analysis time interval, or the number of reflections, or the
distribution uniformity into a time interval (early, late or RT, nor the energies involved. Only analyses the
total); as the method allows to select bands, it can be said uniformity of the existing reflections, whether many or
that the Global SFDC is the average SFDC of a limited few, in the time range of analysis independently if it is
bandwidth impulse response (IR). The diffusion short or long. There may be a high value of SFDC in a
coefficients SFDC by third octave bands describe the very short time period analyzed (eg.: between 80 ms and
uniformity of reflection’s temporal distribution and 120 ms), and a very low value of SFDC in a long time
reflection’s amplitude control into a time interval of a analysis interval (eg.: from 80 ms to 2500 ms), and vice
filtered impulse response using normalized band pass versa.
filters according to IEC 61260 [7]. Depending on each SFDC is intended to be applied initially on a
case both tsplit and Band Energy Level, BEL, can be monaural IR, captured with an omnidirectional
defined. The BEL is the fallen amplitude in dB from the measurement microphone. In addition, the IR’s
maximum of the IR, which stablishes the end of the registering method should bring the largest signal to
analysis time for the SFDC. To evaluate the amplitude noise level (S/N) possible, to obtain the largest result’s
control of diffusing surfaces and the degree of uniformity bandwidth. That’s why is recommended to use a very
in the temporal distribution of reflections, the direct long logarithmic sound sweep, LSS, as stimulus signal,
sound should have to be avoided from the calculus of assuring the best possible environmental conditions
SFDC, by defining the end of it at t direct. A general scheme inside the room to measure (avoiding impulsive noises,
of time and amplitude limits needed for the calculation of and reducing background stationary noise as possible).
SFDC is shown in Fig. 1. It is recommended to use tsplit To get results between 0 and 1, extreme “absolute
of 80 ms for large and reverberant rooms such as concert values” were needed to compare with. As the theoretical
halls, theaters, and auditoriums. A tsplit of 30 ms is condition of total diffusion is not achievable [9], real
recommended for rooms with short reverberation time, maximum values were found.
RT, such as control rooms and studios. The whole IR Results of SFDC processing the same IR were found
analysis is recommended if no early and/ or late to vary mainly in function of selected tsplit, tdirect and
information is needed, for example at reverberation BEL [dB]. Finally, the only variable that depends on the
chambers, control rooms and recording studios. quality of the registered IR is BEL. It was also found that
Nevertheless, to proceed to the comparison of different exists a BEL [dB] interval and a frequency bandwidth for
results of SFDC by bands under identical conditions of which most of the SFDC band’s results remain constant.
analysis, the SFDC results should be accompanied by the
time interval analyzed, tsplit and BEL used (Examples:
4. DESCRIPTOR
SFDC, late, 50ms, 23 dB; SFDC, early, 80ms, 18 dB),
and for the Global SFDC, the included averaged bands
4.1. Calculation
(Example: 250 Hz 10 KHz).
The SFDC is the multiplication of the degree of
uniformity of reflection’s temporal distribution and the
reflection’s amplitude control:
The larger k found in an IR, the less amplitude 4.2. The absolute values
control, due the existence of discrete high-amplitude In the way to find the extreme (maximum and
reflections; thus a low k value means high reflection’s minimum) values of dist_coef and k, respectively, it was
energy control or “uniformity” of amplitudes into the found that SFDC’s results for some real spaces showed
analysed IR. It is important to mention that the whole more diffusion than computer models with maximum
average of h2 always is 1. scattering coefficient on their surfaces. So the absolute
Both dist_coef and k may take values from 0 to values were found from real spaces with well registered
infinite, so to get results of SFDC from 0 to 1, this is to monaural, omnidirectional, using as sound stimulus a log
say absolute results of SFDCs, maximum absolute values sine sweep, and full bandwidth, IRs.
of temporal distribution, dist_coef, and minimum To assure IRs were recorded in the far field of
absolute k values, by bands, were needed to compare diffusers reflections, big halls were chosen, all of them
with. Equation 6 shows the relative result of SFDC, and with Schroeder frequency [10] less than 100 Hz.
Eq. 7 shows the way those values are converted to After analyzing more than 300 IRs, only 81 were
absolute ones. selected from 26 Halls.
Results show the different late diffusion As can be seen, effects of the diffuser are measured
characteristics of the sound field at each position. In this in terms of differential SFDC. Positive values of
case, this information is useful to compare seat positions Differential SFDC mean diffusion was increased.
and / or determine the precise quantities of diffusion a Negative values of Normalized standard Deviation mean
project needs. the spatial uniformity of the sound field was increased.
6.2. Measuring scattering effects in a sound This way, a statistical measure of the footprint of the
field scattering surface could be made in a similar way as the
ISO absorption coefficient measurement in a
With the intention to observe the SFDC variations in reverberation room [15], containing the information of
a sound field while introducing a circular diffuser in it, amplitude control, degree of uniformity of reflections’
initial and final values were measured in 4 microphone temporal distribution and spatial uniformity produced by
positions inside the room. Results applying (9) and (10) the unit.
show the variations of the SFDC and the Normalized
standard deviation of the mapped points in figure 6.
8. REFERENCES
[1] Jeon, J. Y., Kim, Y. H., Vorländer, M. “Counting
local peaks in impulse responses for evaluation of the
in‐situ diffusion in concert halls”. Journal of the
Acoustical Society of America, 129, 2501. (2011).
[2] Kim, Y. H., Yoo, H. J., Jeon, J. Y. “Perception of
scattered sounds in rectangular concert halls”. ISRA
2010, Melbourne, Australia. 2010.
[3] Kim, Y. H. “Evaluation of Wall Diffusers for the
Acoustical Design of Concert Halls”. Doctoral
dissertation. Department of Sustainable Architectural
Engineering Graduate School, Hanyang University.
(2011).
[4] Jeon, J. Y., Kim, Y. H. “Investigation of sound
diffusion characteristics using scale models in
concert halls”. NAG/DAGA International
Conference on Acoustics 2009. The Netherlands.
(2009).
[5] Hanyu, T. “Analysis method for estimating
diffuseness of sound fields by using decay-cancelled
impulse response”. ISRA 2013. Canada. (2013).
[6] Randall, K. E. “The measurement of sound diffusion
index n small rooms”. Research department Report
No. 1969/16. UDC 534.84. BBC. (1969).
Figure 7. Diffuser tested with differential 7 IEC 61260 ED. 1.0 B:1995. (1995).
SFDC analysis. [8] Rose, P. Forensic Speaker identification. Ch. 8.
Taylor & Francis. (2002).
It can be seen that, whatever the case, the possibility [9] Kutruff, H. “Room Acoustics”. Ch. 8. 4th Edition.
of quantifying the diffusion of a sound field enables to Spon Press. (2000).
relate the diffusion and scattering coefficients with direct [10] Skålevik, M. “Schroeder Frequency Revisited”.
effects on the sound field, in a similar way the absorption Akutek. Forum Acusticum. 2011.
coefficient of surfaces are related to changes in the [11] https://acousticengineering.wordpress.com
reverberation time of a sound field.
[12] De Lorenzo, F., calabró, M. “Kurtosis: A statistical
approach to identify defect in rolling bearings”.
2nd International Conference on Marine Research
and Transportation. Italy. 2007.
7. ACKNOWLEDGEMENTS [13] Bidondo, A., Vazquez, S., Vazquez, J., Heinze, G.,
Arouxet, M., Saavedra, A. “Sound Field Diffusion
This research was supported by UNTREF University, Coefficient: Development & Definitions”. ICSV 22.
and their main authorities, Jozami, A. and Kaufmann, M., Italy. 2015
Rector and Vice Rector respectively. Also we have to [14] Bidondo, A., Vazquez, S., Vazquez, J., Heinze, G.,
thank the Environmental Acoustics Group, ITEFI, CSIC, Arouxet, M., Saavedra, A. “Sound Field Diffusion
from Spain, Marzin Zastawnik from AGH University, Coefficient: Absolute values”. ICSV 22. Italy. 2015
Yong Hee Kim from Hanyang University, Urquiza, N., [15] ISO 354:2003. Acoustics – Measurement of sound
Loria, J. M., Olivera, S. and Luquet, E., students from absorption in a reverberation room. International
UNTREF, http://isophonics.net/, and Organization for Standardization. 2003.
http://www.openairlib.net/ who provided impulse
responses or measurement facilities.