The 33 rd International Congress and Exposition on Noise Control Engineering DAAD: A New Software for Architectural Acoustic Design Enis Ozgur a, Feridun Ozis b, Adil Alpkocak a a Dokuz Eylul University, Department of Computer Engineering, 35100 Izmir, Turkey b Dokuz Eylul University, Department of Musicology, 35320 Izmir, Turkey a,b {enis.ozgur,feridun.ozis,adil.alpkocak}@deu.edu.tr Abstract [456], this paper summarizes the ongoing project of DAAD which stands for Dokuz Eylul Architectural Acoustic Design software. The demand for the use of computers in acoustic design has led to the rapid growth of the theory in this area and the increase of room acoustic software. DAAD, dealing with the exact calculation of the impulse response, aims to design an amendable acoustic simulation program by using basic modeling methods. The project focuses on to calculate acoustic parameters of large scale spaces like concert halls and auditorium that are given by ISO as well as 3D modeling and visualization of the building. The project currently uses ray tracing method to achieve impulse response of the volume for the receiver and source points chosen by the user, and wave signals with room effects can also be heard. Moreover, DAAD is also capable to obtain impulse responses for the entire space by dividing the space into small cubes. This new method also helps to visualize the entire space respect to any acoustic parameters. DAAD calculates receiver s size, which can be a spherical or cube shape, according to space volume and quantity of rays. Consequently, the program features the main characteristics of an acoustic simulation program that makes possible to calculate parameters by inserting receivers and sources or, dividing the space in to cubes. The results obtained from simulation experimentation using DAAD has been compared to some benchmarks of well-known concert halls such as Elmia, and has been found very close to real measurements. All the details of method used in DAAD end experimentation result is presented in detail. INTODUCTION The studies on acoustical simulation software have a rapid growth last 20 years. The main purpose of this software is to simulate the propagation of the sound computer in order to get an idea about the room acoustic. The process of acoustical simulation software has three main parts; the first part is source conception which deals with how the sound is emitted. Second part is the modeling method of room which is the most important part of the sound simulation software and is also important for the correct parameters. The third part is modeling of receiver. 1/8
Last two decades, the simulation software is developed by in many company and university. The most important are ODEON [1], CATT Acoustics [2], DIVA [3], CAA [4]. The methods used in acoustic software are shown in Figure 1 to obtain accurate impulse response. SOUCE Source Modelling adiation of sound Sound Synthesis OOM oom Modelling Modelling Methods ay Based Wave Based ECEIVE Modelling of Audience Modelling of Binaural Hearing Figure 1: Modeling of sound in a room DAAD (Dokuz Eylul Architectural Acoustic Design) group has been established in 2002 to develop a new software called DAAD [5] [6], mainly, to calculate the acoustical parameters defined by ISO [7] for large room. System provides user friendly interfaces to design rooms in 3D environments and to locate the receiver and the sound source at desired positions. And, it determines the characteristics of the room by using ray tracing method. Finally, system allows users to listen to the sound affected by room acoustics at receiver position. The system was developed by using Microsoft Visual C++ 6.0, OpenGL 1.2, and Intel C++ Compiler under Windows NT platform. The remainder of the paper is organized as follow. Section 1 describes room modeling tool and the major capabilities and Section 2 explains source and receiver modeling. Acoustic modeling method is explained in section 3. Section 4 presents a new model called receiverless model Section 5 gives the results of our experimentation. And the last section concludes the paper and gives a look to the future studies on this subject. 1 OOM MODELLING TOOL The oom Modeling Tool (MT) of DAAD provides the complete solution for architectural acoustics design in order to acousticians modify the room object and its materials easily. It also provides interfaces to locate all the objects (such as walls, surfaces etc.) and the absorption coefficients are easily set. For this reason, MT is an integrated part of DAAD software and it is more flexible than many of the other 3D drawing software. As shown in Figure 2, it has some outstanding features which are unique to MT, such as opening more than one room and to view the room from many cameras simultaneously. MT also has an ability of reading DXF file format, which make is possible to import existing architectural drawings. 2/8
Figure 2: DAAD oom Modeling Tools 2 SOUCE AND ECEIVE MODELING The source and receiver modeling is an important issue in room acoustic where improper receiver modeling eventually leads to multiple detection, diminished detection or wrong detection. The sources and receivers for acoustical measurements may have different polar diagrams. DAAD uses point sources. Up to date, several receiver modeling has been proposed. Because only spherical receivers allow omni-directional characteristics in sound rays, it has been implemented in most ray tracing models. Furthermore, It has been proved by Lehnert [12], that using a fixed sized receiver model produces a system error. In a recent study, Xiangyang et.al. [8] recommended a model for spherical receiver and shown that it is superior than any previous model on this subject. DAAD uses the model presented by Xiangyang et.al [8] for receivers, where the receiver volume is a function of the number of rays, the distance between the source and the receiver and the volume of the room. This method is explained in Equation 1-2. V reciever k d S 4 N = (1) where V reciever is volume of receiver, d is distance between sound and receiver, N is the number of ray and V room is the volume of room. k = log ( Voom) (2) 10 3/8
ECEIVE SOUCE Figure 3: DAAD Source and receive location 3 AY TACING AND CALCULATION OF IMPULSE ESPONSE Impulse response has all the acoustical clues of a room in itself. In acoustical simulation software the different modeling methods are used for obtaining impulse response of the room as shown in Table1. As the ray based methods are successful in modeling of high frequency sound waves, waves based methods are successful in modeling of low frequency sound. But wave based methods are not preferred because of the cost of computation for large rooms in computers, so DAAD uses ray tracing method. Table 1: Methods of calculation impulse response. ay-based Methods ay Tracing Beam Tracing Hybrid Method Wave-based Finite Element Boundary Element In DAAD, the room simulated is assumed a Linear Time Invariant System. As shown in Figure 4, the fix number of rays emitted in equal angles spherically and reflects from surfaces of the room as Snell laws. The numbers of rays and reflections are set by user. 4/8
α eceiver α Source S Figure 4: ay Tracing Method It is assumed that they carry Dirac function [9], when the rays are emitted from the source. This Dirac function scales according to the absorption coefficient of the surfaces reflected and shifts as the time from the source to receiver. For a single ray, impulse response of the room for a given receiver at location and for j th octave band on time t, h j (t), is calculated as shown in Equation 3: d i h j ( t) = ( g ij ) hs t (3) i= S i= S c where source point is S, the distance from the source to the receiver is d, speed of sound is c, impulse response at location S is h s, the absorption coefficients of i th surface for j th octave band is g ij. So, the impulse response of the room for the j th octave band, h j, is represented as in Equation 4: h j = {h j1,h j2,,h jm } (4) where m is the number of samples and an arbitrary entry of h j is defined follows: h jk = max[h j (k)] (5) Equation 3 is calculated for each octave band since each material may have different absorption coefficient for different octave band. The resulting impulse responses for eight octave bands, h, is represented as follows: { h h h } h = (6) 1, 2,..., 8 The impulse responses are transformed into frequency domain and represented as follows: { w w w } w = (7) 1, 2,..., 8 Finally, we have frequency responses for each octave band. It needs to be summed together as in Equation 8, in order to get a final impulse response of the room. 8 ( w) = f ( ) h (8) w j j= 1 5/8
where f is defined as a filter function for octave bands to trim the frequencies from respective frequency response, b min and b max are lower and upper limits of desired octave band, respectively. x if bmin j < x < bmax j f ( x) = (9) 0 if not As a result, room frequency response, h r (w) is obtained. The whole process of calculation and auralization are shown in Figure 6 as a block diagram. Figure 6: Total impulse responses of room and auralization In DAAD, to hear the sound of a room modeled, the given sound wave is convolved with impulse response of the room, h (t). The resulting sound signal is that how to hear the sound which is emitted from the source at the receiver. The aurialization process is shown in Figure 6. 4 ECEIVELESS MODEL In [10], indel states that it can be extremely useful for the acousticians to see a mapping of spatial distribution of acoustical weak spots can easily localized and appropriate countermeasures can be taken. Beyond this idea, it is also useful to see a volumetric display of a room respect to some acoustical parameters calculated. This helps us to realize a walkthrough in room, or even flythrough for potential applications of acoustic simulations. In this receiverless model, we propose a model where no specific location is required. This is because, only a small modification on the single receiver model provides us to render whole volume for acoustical parameters. This is easily done by dividing the whole volume into equal-sized cubes bounded by a spherical receiver located in the center of the each cube. Computationally, the most costly fraction of the ray tracing model with a single receiver is the part given in Equation 3. However, in this equation number of receiver is not a factor of h (t). It means, the all h j (t) for all receivers/cubes where the ray pass-trough simply calculated with 6/8
voxelization algorithm while h j (t) is calculated for a single ray. Moreover, this modification, filling the volume with identical receivers does not add too much overhead computationally. In other say, the computational cost is almost same with respect to single receiver model. This model leads many open research problems such as optimum grid size, or unit receiver size calculation. The number of receivers or grid size is defined by user and, then DAAD automatically locates the receivers as shown in Figure 7. DAAD considers the room as 3D grid containing a receiver in each cell. Figure 7: Grid Model of Elmia 5 PEFOMANCE EVALUATION The results of DAAD are compared with the international ound obin 2 (II) which is about the concert hall Elmia in Sweden [11]. Table 2 shows the comparison between DAAD results and the average results of -II. The comparison is for T30, EDT for S1 and 1 position. The rows labeled -II indicate the average values for six octave band and its standard deviation of each value has been shown as ±. Table 2: DAAD results comparison with -II Hz 125 250 500 1000 2000 4000 T30/s EDT/s II 1.68 ± 0.3 1.95 ± 0.4 2.08 ± 0.4 2.11 ± 0.4 1.97 ± 0.3 1.67 ± 0.2 DAAD 1.51 1.55 1.58 1.88 1.70 1.68 II 1.63 ± 0.4 1.87 ± 0.4 2.02 ± 0.4 2.03 ± 0.4 1.89 ± 0.3 1.6 ± 0.4 DAAD 1.48 1.55 1.58 1.69 1.60 1.55 7/8
The experimentation results show that DAAD is better than many participants to II even though we have not implemented the diffraction and scattering effect yet. The readers who want to get the complete set of results for Elmia may refer to web page at http://pamir.cs.deu.edu.tr/daad. 7 CONCULUSION In this study, we have presented the details of the ongoing project of DAAD which stands for Dokuz Eylul Architectural Acoustic Design software. DAAD, dealing with the exact calculation of the impulse response, aims to design an amendable acoustic simulation program by using basic modeling methods. DAAD is capable to calculate acoustical parameters given by ISO as well as 3D modeling and visualization of the building. In this project, we have used the ray tracing method to achieve impulse response of the volume for the receiver and source points chosen by the user, and wave signals with room effects can also be heard. Moreover, DAAD is also capable to obtain impulse responses for the entire space by dividing the space into small cubes. This new method also helps to visualize the entire space respect to any acoustic parameters. Consequently, the program features the main characteristics of an acoustic simulation program that makes possible to calculate parameters by inserting receivers and sources or, dividing the space in to cubes. The results obtained from simulation experimentation using DAAD has been compared to some benchmarks of well-known concert halls such as Elmia, and has been found very close to real measurements. All the details of method used in DAAD end experimentation result is presented in detail. In longer term, we expect the DAAD project to lead us into new research in many dimensions, including scattering and diffraction effects, new visualization techniques and applications of acoustics and more. EFEENCES [1] G. M. Naylor, ODEON Another hybrid room acoustical model, Applied Acoustics, Volume 38, Issues 2-4, pp.131-143, 1993. [2] Dalenbäck, B.-I., CATT-Acoustic: Image source modelling augmented by ray tracing and diffuse reflections, Applied Acoustics, Volume 38, Issues 2-4, pp.350, 1993. [3] DIVA Web Site, http://www.rhintek.com/cara/, 2003. [4] CAA Web Site, http://www.bksv.com/, 2003. [5] Özgür Enis, Design and Development of an Architectural Acoustic Design Software, MSc thesis, Dokuz Eylül University, Department of Computer Science and Engineering,200. [6] Sarıgül Yavuz, Sound Modeling for oom Acoustic, MSc thesis, Dokuz Eylül University, Department of Computer Science and Engineering, 2003. [7] ISO 3382, Measurement of the reverberation time of rooms with reference to other acoustical parameters, 2003. [8] Zeng Xiangyang, Chen Ke an, Sun Jincai, On the accuracy of the ray-tracing algorithms based on various sound receiver models, Applied Acoustics, 64, 433 441,2003. [9] Arnost V., Discrete Simulation of Sound Wave Propagation, MOSIS 2000 Proceedings, 241-246, 2000 [10] indel, J.H., The Use of Computer Modeling in oom Acoustics, Journal of Vibroengineering, No.3(4), 2000. [11] ound obin Web Site, http://www.ptb.de, 2003. [12] Lehnert Hilmar, Systematic Errors of the ray-tracing algorithm, Algorithm Acoustics, Volume 38, pp. Issues 2-4, 207-221, 1992. 8/8