A A A Performance optimisation of small reverberant room with hanging diffusers Patrick Bouche 1 , Kevin Verdière 2 , Simon Campeau 3 Mecanum Inc 2444 rue Bonin, J1K 1C4, Sherbrooke, QC, CA ABSTRACT A small reverberant room can be an efficient and economic tool to provide fast diffuse field sound absorption measurements of homogeneous sound absorbers or more complex structures, and useful for developing, testing, or evaluating specification requirements in many industrial fields. However, small reverberant rooms are known to have some diffusivity issues in lower and middle frequencies, leading to unsuitable levels of consistency, reliability, and repeatability. This paper presents the re- sults of a study made to improve the measurement performance of a 5.7-m³ reverberant room under its Schroeder frequency (around 1275 Hz) by adding hanging diffusers. Using Ray-Tracing method, a numerical parametric study was done to estimate sound absorption of various samples by varying the number, the position, and the orientation of these diffusers based on ASTM C423 and E795 stand- ards. Moreover, extra simulation has been performed to evaluate the effect of sample size on sound absorption consistency in function of frequency. Following these prescriptions, an experimental study was done to confirm these improvements on frequency dependent sound absorption and single rating numbers such as NRC (Noise Reduction Coefficient) and SAA (Sound Absorption Average). The re- sults show that for this room, low frequency performance is significantly increased by using 4 well-placed hanging diffusers and a sufficient sample size. 1. INTRODUCTION Reverberant rooms are conventional equipment used by researchers and manufacturers to measure the diffuse field absorption coefficient of a large range of materials, from homogeneous porous struc- tures to complex combinations of different elements. This indicator is used to predict acoustic per- formance of these materials in many fields, like car or building industries. This measurement instru- ment follows multiple industrial international standards: the ASTM C423 [1] or ISO 354 [2] standards for rooms above 200 m 3 , and the SAE J2883 [3], that adds specific limitations to the two previous standards for the case of small rooms, with a volume between 6 and 25 m 3 . In this study, a 5.7-m 3 small reverberant room is considered and used (Figure 1). Its interior volume has a rectangular shape of 2.5 by 1.3 by 1.7 m and is made of six panels with a high acoustic reflec- tivity. The acoustic field is generated by four uncorrelated loudspeakers positioned at each upper 1 patrick.bouche@mecanum.com 2 kevin.verdiere@mecanum.com 3 simon.campeau@mecanum.com worm 2022 corner. The sound pressure level is measured by four microphones placed on the surface of an imag- inary 0.3-m radius sphere placed in the center of the room. Figure 1: Photo of the small reverberant room In 1962, Manfred Schroeder has presented a limiting frequency separating the modal and diffuse regions using a statistical approach. The cut-off frequency of a room, named Schroeder frequency, is defined by: worm 2022 𝑓 𝑐 = 2000 ඨ 𝑅𝑇 60 𝑉 ≅1275 𝐻𝑧 (1) where 𝑅𝑇 60 is the reverberation time, in seconds, for a 60-dB decay and 𝑉 is the room’s volume. Most international standards state that a good and reproductible sound absorption measurement is linked to a high sound field diffusivity in the reverberant room. Therefore, with such a high Schroeder frequency, low frequency absorption measurements in small rooms can’t be reliable without improv- ing the sound diffusiveness. Following the SAE J2883 prescriptions [3], with a volume of 5.7 m 3 and a minimal distance between 2 microphones of 0.45 m, the cut-off frequency of this reverberant room is lowered to 500 Hz. Considering this low frequency limitation, this paper presents an approach to improve the reliabil- ity and repeatability of the measurement of sound absorption in small reverberant room by adding hanging diffusers, as described in ASTM C423 [1] and ISO 354 [2]. To achieve this, a numerical investigation is carried out to improve the sound field diffusivity on the whole sound spectrum, with a special focus on the viability of measurements under the cut-off frequency and down to 250 Hz. This investigation is then followed by experimental validation. 2. PURPOSE AND METHODOLOGY To reach a high degree of sound field isotropy, ASTM C423 [1] in section 7.4 and appendix X1 and ISO 354 [2] in appendix A suggest adding a sufficient number of diffusers (i.e. increase the total surface of diffusers) until the absorption coefficient for a test specimen is maximized ( 𝛼 𝑚𝑎𝑥 ). There- fore, this study aims to find the diffuser configuration that produces the maximum absorption to create an approximately diffuse sound field. 2.1. PROTOCOL According to theses standards, stationary and rotating hanging diffusers are strongly recommended to increase diffusivity and achieve an acceptable diffuseness. It’s also strongly recommended to give random orientations and positions to those panels inside the reverberant room. A global optimisation is done using numerical simulation with the Ray Tracing method by increasing the surface area of diffusers for different test specimens of various dimensions and acoustics properties. Due to the small size of the room, to evaluate the improvement of diffuseness, the diffuser area is changed by adding an increasing the number of the same base diffuser, from 0 to 5, instead of the 5-m² step increase proposed by the standards. The sound absorption coefficient of test specimen is given by: 𝛼 𝑆 ≅ 55.3 𝑉 𝑐 0 𝑆 1 − 1 ൨ (2) 𝑇 1 𝑇 0 where 𝑐 0 is the speed of sound, 𝑉 is the room volume, 𝑆 is the surface of test specimen, 𝑇 0 is the reverberation time without test specimen, and 𝑇 1 is the reverberation time with test specimen. Two main indicators are used to evaluate the diffusivity of the sound field and the reproductivity and repeatability of the measurement. They are: the average of the sound absorption coefficient ( 𝛼 ) for different floor positions for a specimen the normalized standard deviation of the sound absorption coefficient ( 𝜎 𝑛 ) The normalized standard deviation is the standard deviation divided by the mean. Using this sta- tistic makes it more appropriate to compare the spread of the distribution of one variable with large mean and standard deviation with the distribution of another variable with small mean and standard deviation. As an example, a normalized standard deviation of 10% can give a value of 1±0.1 or a value of 0.2±0.02. 2.2. HANGING DIFFUSERS The hanging diffusers used are 2.3-mm thick rectangular aluminum sheets of 0.9 m by 0.5 m covered by a viscoelastic material. The total surface area is 0.9 m 2 (both sides are counted). These diffusers are shaped in a S-curved structure. Each diffuser is attached to the ceiling by three metallic cables and magnetic anchors. 3. NUMERICAL SIMULATIONS 3.1. RAY TRACING Ray tracing is a common method in physics for calculating the path of any kind of waves through a 3D environment which may contains various domains and obstacles. The propagation of waves depends on the properties of the regions they pass through. Under specific circumstances, such as worm 2022 encountering an obstacle, waves may change direction, reflect, or be absorbed. The problem is solved by releasing a discrete number of beams or rays through a complex medium. One specific application of the Ray Tracing method that can be applied to the acoustic field is the Particle Tracing. In that special case, each emitted ray is considered as a carrier of acoustic energy travelling inside the reverberant room at the speed of sound. After each reflection on a surface, the energy carried by the ray is reduced according to its absorption coefficient. The global energy decay of the reverberant room can be given by displaying the energy of all particles in function of the time. Using this result, the reverberation time ( 𝑅𝑇 60 ) can be calculated quickly and accurately. However, this method is highly dependent of the position of the sources, the walls and any obstacle in the room and their respective acoustic properties. As previously investigated by Toyada et al. [4], using this method sound absorption coefficient measurement in large reverberant room can be effectively im- proved by adding hanging diffusers to increase the global diffusivity. To predict the influence of the hanging diffusers, a list of different viable combinations of positions and orientations is generated randomly. A viable combination is obtained when no spatial interference occurs between diffusers. Then, the testing room is modeled using the ray tracing module included in COMSOL Multiphysics software [5]. The four loudspeakers are represented by hemispherical sound sources generating pulses of 4000 rays. The main direction of each hemispheric emitter is aligned with the orientation of the loudspeakers in the room. Calculations are made with a step of 0.1 s from 0 to 3 s. Four spherical receivers of 0.01-m radius are placed at the locations of the micro- phones to calculate the impulse response and determine the energy decay in function of time. This model allows to simulate the actual test method. It evaluates the sound absorption coefficient of each test specimen measured under every predefined combination of 0 to 5 diffusers. Each test specimen is rectangular and placed on the floor. Figure 2 presents a model in the case of 4 diffusers, the loud- speakers and the receivers are represented by respectively yellow and red dot. worm 2022 Figure 2: View of the Ray Tracing Method in the case of four diffusers. The rectangle (in Blue) on the floor corresponds to the test specimen. 3.2. MATERIAL SPECIMENS To simplify the simulation and avoid potential diffraction effect, the tested materials are defined by their sound absorption in third octave bands and directly applied to the rectangular floor surface shown in Figure 2. Three different test specimens ( S 1 , S 2 and S 3 ) are used. Their sound absorption curves are presented in Figure 3. 1.2 1 Absorption coefficient 0.8 0.6 0.4 S1 S2 S3 0.2 0 250 315 400 500 630 800 1000 1250 Third Octave Bands (Hz) Figure 3: Predefined absorption coefficients of the three test specimens in third octave bands (Hz) For each material, test specimens of different areas are generated to analyze the impact of the overall dimensions on the results. The dimensions of each rectangular surface area are presented in Table . Table 1: Dimensions L (m) W (m) S (m 2 ) W/L 1 0.6 0.6 0.36 1 2 0.6 0.8 0.48 1.33 3 0.8 0.8 0.64 1 4 0.8 1 0.8 1.25 5 0.8 1.25 1 1.56 6 1 1.25 1.25 1.25 Each test specimen is randomly placed on the floor following three rules: - The sides of the rectangle should not be parallel to the walls - The center of the rectangle must not be in the center of the floor - A minimal distance of 0.1 m is kept between the border of the test specimen and the walls. worm 2022 3.3. EFFECTS OF HANGING DIFFUSERS NUMBER ON THE ABSORPTION COEFFI- CIENT The results of this optimisation process led to the conclusion that 4 hanging diffusers are optimal to obtain a Sabine absorption coefficient equal or higher than the absorption of the test specimen for the most promising configurations. Results presented below are given from specimen 𝑆 2 . Same con- clusions are obtained with 𝑆 1 and 𝑆 3 . Because this paper aims to improve low frequency measurement (i.e., diffusiveness), results are presented below the Schroeder frequency. Figure 4 presents the simulated Sabine sound absorption coefficient of the test specimen 𝑆 2 ob- tained with and without 4 optimally configured hanging diffusers, compared to its predefined value. Figure 5 presents the simulated sound absorption at 400 and 800 Hz in function of the number of diffusers. It shows an increase of the absorption value with the number of diffusers, with a stagnation beyond 4. The 4-diffusers sound absorption coefficient appears to be higher than the predefined value of the test specimen, especially for high absorption values. This phenomenon has already been shown in previous scientific publications [4, 6, 7, 8] both in numerical simulations and experimental tests. Apart from some software errors, these higher values might be induced by the fact that a test specimen with high absorption can influence the diffusivity and the mean free path of the sound waves inside the reverberant room. 1 1 0.9 0.9 0.8 Absorption coefficient Absorption coefficient 0.8 0.7 0.6 0.7 0.5 0.6 0.4 S2 -Defined 0.3 0.5 S2 - 0 Diffuser 0.2 0.4 S2 - 4 Diffusers 0.1 400-Hz 800-Hz 0.3 0 0 1 2 3 4 5 250 315 400 500 630 800 1000 1250 Third Octave Bands (Hz) Number of diffusers Figure 4: Simulated absorption coefficient of Figure 5: Evolution of the absorption coeffi- cient at 400 and 800-Hz in function of the test specimen S 2 for 0 and 4 diffusers number of diffusers 3.4. EFFECTS OF HANGING DIFFUSERS POSITION ON THE ABSORPTION COEFFI- CIENT In the previous section, the optimal number of diffusers has been found to be 4. In this part, we will discuss about their relative position in the reverberant room. Simulations have been made to highlight the effect of the orientation of the diffuser itself and the global position of the 4 diffusers. Intermediate results are not presented in this paper. Some diffuser configurations didn’t affect the sound absorption coefficient for some or every third octave bands, showing that the effect of the hanging diffusers depends on their global position (main parameter) in worm 2022 the reverberant room. The vertical position of the 4 diffusers has the most influence. In conclusion, this study part will focus on finding an optimised vertical position for this reverberant room configu- ration using its specific setting of speakers and microphones positions. The influence of diffusers vertical position on the simulated sound absorption coefficient of the test specimen 𝑆 2 (1-m 2 area) is presented in Figure 6. To evaluate this effect, a set of 4 randomly oriented diffusers contained in a 0.5 m vertical space is moved at 3 different heights, 0, 0.25 and 0.5 m from room ceiling, forming zones A, B and C. Simulated results presented in Figure 6 for each case is an average of several random configurations of 4 diffusers in each defined zone. The sound absorption at 400 and 800-Hz is also presented in function of the number of diffusers for each zone in Figure 7 and Figure 8 . Defined No Diffuser Zone A Zone B Zone C 1 0.9 0.8 worm 2022 Absorption coefficient 0.7 0.6 0.5 0.4 0.3 0.2 0.1 250 315 400 500 630 800 1000 1250 Third Octave Bands (Hz) Figure 6: Simulated absorption coefficient of the test specimen S 2 for different diffusers location 0.65 1 Zone A Zone B 0.95 0.6 Zone C Absorption coefficient 0.9 Absorption coefficient 0.55 0.85 0.8 0.5 0.75 0.45 0.7 0.4 0.65 0.6 0.35 0 1 2 3 4 5 0 1 2 3 4 5 Number of diffusers Number of diffusers Figure 7: Evolution of the absorption at 400- Figure 8: Evolution of the absorption at 800- Hz in function of the diffuser’s location Hz in function of the diffuser’s location The best results are obtained when the diffusers are contained in the zone B, between 0.25 and 0.75 m from the ceiling, leading to a global increase of the absorption on all octave bands. Zone A results show the same increase under 500 Hz, but an insufficient increase above compared to the predefined value. On the opposite, zone C results show an insufficient increase of the absorption coefficient under 500 Hz, but a larger increase over 630 Hz. The fact that the absorption coefficient varies in function of the position of the diffuser may indicate the presence of a near field system in which the diffusivity pattern of each diffuser is influenced by nearby elements. Moreover, one can observe that the number of diffusers affects differently the measurement depending on their vertical location. 3.5. EFFECT OF HANGING DIFFUSERS ON CONSISTENCY AND REPEATABILITY Another key point of this study was to evaluate the effect of adding hanging diffusers on the con- sistency and the repeatability of the measurement. As mentioned earlier in 4.2, each test specimen has been simulated with different surface areas and multiple locations on the floor of the room. The mean normalized standard deviation of the sound absorption is presented in Figure 9 for sur- face area of 0.64 m 2 and 1 m 2 . This value is given by the average value of the normalized standard deviation of each test specimen in third octave bands for the same surface area. These results show a large reduction of the normalized standard deviation with 4 hanging diffusers in the optimized loca- tion from 12.6% to 5.3% and 11.4% to 4.1% at 630 Hz respectively for a surface area of 0.64 m 2 and 1 m 2 . worm 2022 45% 45% 40% 40% Normalized standard deviation Normalized standard deviation 35% 35% 30% 30% 25% 25% 20% 20% 15% 15% 10% 10% 5% 5% 0% 0% 250 315 400 500 630 800 1000 1250 250 315 400 500 630 800 1000 1250 Third Octave Bands (Hz) Third Octave Bands (Hz) Figure 9: Normalized standard deviation of the sound absorption coefficient in function of the third octave bands pour sample's surface of 0.64-m 2 (right) and 1-m 2 (left) 4. EXPERIMENTATIONS 4.1. MATERIALS To validate the proposed approach, three acoustic materials are used: a melamine foam, a shoddy- based fibrous and a polyurethane convoluted foam. These three conventional acoustic materials are assumed to be homogeneous and isotropic. The Johnson-Champoux-Allard (JCA) properties [9, 10] of melamine foam are measured with the characterization equipment’s of Mecanum Inc [11]. The density-based shoddy model, developed by J. Manning et al [12], is used to determine the JCA prop- erties of the shoddy-based fibrous. All characterized and determined properties for the melamine foam and the shoddy-based material are presented in Table . The polyurethane foam is only used to evaluate the effect of the size of the test specimen and haven’t been characterized. Table 2: JCA properties Melamine Shoddy-based Foam fibrous Thickness (m) 0.05 0.025 Density (kg/m 3 ) 8.89 42.7 Porosity 0.989 0.964 Resistivity (N.s/m 4 ) 11132 9905 Tortuosity 1.02 1 Viscous length (m) 1.127e-3 8.351e-5 Thermal length (m) 3.046e-3 1.351e-4 4.2. IMPLEMENTATION OF THE PRESCRIPTIONS GIVEN BY NUMERICAL SIMU- LATIONS worm 2022 To confirm prescriptions given by the ray tracing simulation, hanging diffusers are added one by one in the best configuration until the maximum absorption coefficient is reached. The experimental study is done as follow. Figure 10: View of 4 diffusers installed in the small reverberant room In accordance with ASTM C423 [1] standard appendix X1, sound absorption measurements were done on a 25-mm thick Shoddy-based fibrous material and a 50-mm thick melamine foam. The area of the test specimens is 1-m 2 (0.8-m by 1.25m for the shoddy and 1 m by 1 m for the melamine foam). A first measurement was made without diffusers for 5 random sample locations on the floor. This measurement process was repeated four times, adding one diffuser at each iteration. The edges of the samples are sealed with tape and a wooden frame to avoid any leaks or edge effects. This set-up refers to a type A-Mounting defined the ASTM E795 standard [13]. Figure 11 and Figure 12 shows the evolution of calculated average absorption coefficient and nor- malized standard deviation with the augmentation of diffusers for the two specimens tested, in third octave bands between 250 and 10,000 Hz. worm 2022 1.2 16% 0 Diffuser 1 Diffuser 14% 1 2 Diffusers Normalized standard deviation 12% Absorption coefficient 3 Diffusers 0.8 10% 4 Diffusers 8% 0.6 6% 0.4 4% 0.2 2% 0% 0 100 1000 10000 100 1000 10000 Third Octave Bands (Hz) Third Octave Bands (Hz) Figure 11: Average (left) and normalized standard deviation (right) of the absorption coefficient of a 25-mm thick shoddy-based material in function of the number of diffusers in the reverberant room 1.2 12% 0 Diffuser 1.1 1 Diffuser 10% 1 2 Diffusers Normalized standard deviation Absorption coefficient 0.9 3 Diffusers 8% 4 Diffusers 0.8 0.7 6% 0.6 4% 0.5 0.4 2% 0.3 0.2 0% 100 1000 10000 100 1000 10000 Third Octave Bands (Hz) Third Octave Bands (Hz) Figure 12: Average (left) and normalized standard deviation (right) of the absorption coefficient of a 50-mm thick melamine foam in function of the number of diffusers in the reverberant room Both results show that the measured sound absorption coefficient increases progressively with the number of hanging diffusers and seems to stagnate between 3 and 4 diffusers. In accordance with the numerical simulation, these measurements induces that an acceptable diffusion has been reached. Results also show that the normalized standard deviation decreases with the number of diffusers. With 4 diffusers, the normalized standard deviation seems to be lower for higher absorption values at low frequencies. As an additional comparison, the measured sound absorption coefficients of both materials are compared to a simulated absorption coefficient under a diffuse field with Mecanum’s NOVA soft- ware [14] in Figure 13. Results show a good agreement between simulation and experimental results on both test specimens with sealed edges. The higher absorption of shoddy over 800 Hz can be ex- plained by the irregular thickness of the sample, creating thin air gaps underneath. worm 2022 1.4 1.3 1.2 1.2 1.1 Absorption coefficient Absorption coefficient 1 1 0.8 0.9 0.8 0.6 0.7 0.4 NOVA FTMM 0.6 Exp with sealed edge 0.2 0.5 Exp with unsealed edge 0 0.4 100 1000 10000 100 1000 10000 Third Octave Bands (Hz) Third Octave Bands (Hz) Figure 13: Comparison of experimental and simulated sound absorption coefficient of the shoddy- based material (left) and the melamine foam (right) 4.3. LOWER FREQUENCY LIMITATIONS AND MINIMAL SPECIMEN SIZE As specified in ASTM C423-17 [1] in section 9, the test specimen should have a recommended area of 6.69 m 2 in a shape of 2.44 by 2.74 m for large reverberant room. It is also mentioned that area under 5.57 m 2 should not be used. This recommendation being unapplicable to the room studied, different specimen sizes were tested to evaluate the effect of their size on the reproducibility and repeatability of the measurement. To achieve different size, combinations of panels of 38 mm thick convoluted polyurethane foam are used. Each panel has a surface of 0.25 m 2 (0.56 by 0.45 m). Four combinations of 2, 3, 4, and 6 panels in respective configuration of 1 by 2, 1 by 3, 2 by 2, and 2 by 3 were tested 5 times each in different random locations on the floor. The edges of the specimens were not sealed due their complex shape. 4 hanging diffusers are installed in the room in their optimised configuration. The average absorption coefficient and normalized standard deviation are shown at Figure 14. The average sound absorption is generally higher for smaller samples, as it can be expected for unsealed edges. More interestingly, the normalized standard deviation decreases as the surface area increases. It goes from 24.6% to 9.2% at 250 Hz or from 8.6% to 2.5% at 500 Hz, meaning that the reproduci- bility and repeatability of measurement is better for larger samples. However, it stays very similar for a test specimen of 1 and 1.5 m 2 on all third octave bands. The minimum target specimen size for a relevant measurement must be 1 m². worm 2022 1.2 25% 0.5-m2 1.1 0.75-m2 Normalized standard deviation 1 20% 1-m2 Absorption coefficient 0.9 1.5-m2 0.8 15% 0.7 0.6 10% 0.5 0.4 5% 0.3 0.2 0% 100 1000 10000 100 1000 10000 Third Octave Bands (Hz) Third Octave Bands (Hz) Figure 14: Average (left) and normalized standard deviation (right) of the absorption coefficient of a polyurethane foam in function of the surface area of the sample 5. CONCLUSION Performance optimisation of a small reverberation room was carried using geometrical acoustics to improve diffusivity with hanging diffusers following standards protocols. Numerical simulations confirmed that adding well positioned diffusers in the room improves the measured sound absorption coefficient. Therefore, with the right configuration, hanging diffusers are an efficient way to increase the sound diffusivity below Schroeder frequency of a room, lowering its effective cut-off frequency. worm 2022 Based on the numerical prescriptions, a similar investigation was done experimentally to validate these conclusions. Results show the same observations on the measured sound absorption. The nor- malised standard deviation of the absorption coefficient of a sample in different locations on the room floor decreases with the addition of diffusers. This is a sign that the diffusers improve the sound field diffusivity, leading to a better repeatability of the measurements at 250 Hz and over. Last experimental investigation shows that under this optimized diffuser configuration, the nor- malised standard deviation of sound absorption decreases when the area of the test specimen in- creases, until it reaches an area of 1 m 2 . Theses results leads to the conclusion that it should be the target sample size to obtain good reproducibility and reliability on sound absorption measurements in this small reverberant room, even under its cut-off frequency. 6. REFERENCES [1] ASTM Standard C423-17 - Standard test method for sound absorption and sound absorption coefficients by the reverberation room method. [2] ISO 354 - In measurement of sound absorption in a reverberant room. [3] SAE J2883 - Laboratory measurement of random incidence sound absorption tests using a small reverberation room. [4] E. Toyada, S. Sakamoto and H. Tachibana, "Effects of room shape and diffusing treatment on the measurement of sound absorption coefficient in a reverberant room," Acoustical Science & Technology, pp. 255-266, 2004. [5] [Online]. Available: https://www.comsol.fr/acoustics-module. [6] T. J. Cox and A. P. D, Acoustic absorbers and diffusers, Taylor & Francis, 2009. [7] C. W. Kosten, "International comparison meaurements in the reverberation room," Acustica, vol. 10, 1960. [8] C. Scrosati, F. Scamoni, M. Depalma and N. Granzotto, "On the diffusion of the sound field in a reverberation room," ICSV 26 International Congress on Sound and Vibration, 2019. [9] D. L. Johnson, J. Koplik and R. Dashen, "Theory of dynamic permeability and tortuosity in fluid-saturated porous media," Journal of Fluid Mechanics, pp. 379-402, 1987. [10] Y. Champoux and J. F. Allard, "Dynamic tortuosity and bulk modulus in air-saturated porous media," Journal of Applied Physics, pp. 1975-1979, 1991. [11] [Online]. Available: https://www.mecanum.com/produits. [12] J. Manning and R. Panneton, "Acoustical model for shoddy-based fiber sound absorbers," Journal of Sound and Vibration, 2013. [13] ASTM Standard E795-16 - Standard practices for mounting test specimens during sound absorption tests. [14] "https://www.mecanum.com/nova," [Online]. [15] M. Vercammen, "Improving the accuracy of sound absorption measurement according to ISO 354," Proceedings of the international symposium on room acoustics ISRA, 2010. worm 2022 Previous Paper 145 of 769 Next