WAVE FIELD SYNTHESIS A PROMISING SPATIAL AUDIO RENDERING CONCEPT. Günther Theile* and Helmut Wittek**

T be published in Jurnal f the Institute f Image Infrmatin and Televisin Engineers, 2007 WAVE FIELD SYNTHESIS A PROMISING SPATIAL AUDIO RENDERING CONCEPT Günther Theile* and Helmut Wittek** * Institut für Rundfunktechnik (IRT), Germany ** Scheps Mikrfne, Germany ABSTRACT Mdern cnvlutin technlgies ffer pssibilities t vercme principle shrtcmings f ludspeaker sterephny by expliting the Wave Field Synthesis (WFS) cncept fr rendering virtual spatial characteristics f sund events. Based n the Huygens principle ludspeaker arrays reprduce a synthetic sund field arund the listener, whereby the dry audi signal is cmbined with measured r mdelled infrmatin abut the rm and the surce s psitin t enable the accurate reprductin f the surce within its acustical envirnment. Hwever, theretical and practical cnstraints f WFS systems limit the rendering accurateness and the perceived spatial audi quality t a certain degree, dependent n characteristic features and technical parameters f the sund field synthesis. Nevertheless recent develpments have already shwn that a number f applicatins culd be pssible in the near future. An attractive example is the synthesis f WFS and sterephny ffering enhanced freedm in sund design as well as imprved quality and mre flexibility in practical playback situatins fr multichannel sund mixes. Furthermre, a nvel applicatin called Binaural Sky cmbines wave field synthesis and binaural techniques. Fcused surces are rendered clse at listener s ears by means f headtracking and a WFS-array abve the listener, frming a virtual headphne. As a result, binaural (e.g. BRS) reprductin can be enjyed withut wearing a headphne and withut any ludspeakers in the listener's field f visin. 1. INTRODUCTION Three psychacustic fundamentally different spatial audi imaging methds shuld be distinguished: (Multichannel) ludspeaker sterephny Binaural recnstructin f the ear input signals Syntheses f the sund field arund the listener All knwn spatial sund systems can be traced back t ne f these methds r can cntain mixed frms theref, whereby certain advantages f the methds are being explited, respectively its disadvantages are avided, dependent n the intended applicatin area. 1.1. Ludspeaker sterephny This is in principle based n the characteristics f lcalizatin in the superimpsed sund field, generated by tw ludspeakers [1]. Directinal imaging is dne in the imaging area between tw adjacent ludspeakers [2]. In the case f 3/2 sterephny, with the help f surrund channels the imaging area between the frnt ludspeakers can be extended. Therefre pssibilities are ffered fr the reprductin f early lateral sund fr imaging f spatial depth as well as reverberatin, in rder t prduce the spatial impressin and the envelpment. Details are described in [3]. 1.2. Binaural recnstructin f the ear input signals The riginal emplyment f this methd is the knwn dummy head sterephny. It is nt intended t reprduce a suitable sund field at the reprductin lcatin. Instead, the effective ear signals at the recrding lcatin are recrded with the assistance f a dummy head and replayed in principle via headphnes. Under ideal circumstances, the reprduced binaural signals are identical t the riginal ear signals that the listener received in the recrding lcatin. In practice it is pssible t reprduce auditry events with excellent realism regarding spatial characteristics and sund clur. 1.3. Synthesis f the sund field arund the listener The third apprach was pursued within the framewrk f the Eurpean Research Prject CARROUSO [4]. It is based n the cncept f Wave Field Synthesis (WFS, develped at the Technical University Delft, refer e.g. [5], [6]), i.e. the representatin f a virtual surce and a virtual rm is achieved by rendering an acustically crrect sund field. The principle f WFS is based n the ludspeaker arrays which generate a cmplete sund field in the listening zne which is identical t an apprpriate real sund event (see Chapter 2). This acustical cunterpart t the ptical hlgraphy is als described as hlphny. The binaural ear input signals that are active fr the auditry event thus arise in a natural way within the sund field, cntrary t dummy head sterephny. 2. WFS PRINCIPLES AND PROPERTIES 2.1. WFS the applicatin f the Huygens Principle The Huygens principle states: If frm a pint S f a hmgeneus istrpic medium a spherical wave is emitted, ne can imagine the prcedure f the individual wave reprductin in that a particle brught int scillatin by external frces, transfers its mvements t its neighburing particles. This prcedure then cntinues symmetrically in all directins and in this way gives cause t a spherical wave... [7]. The Kirchhff-Helmhltz integral, an

applicatin f this fundamental principle, is used as the underlying mathematical frmulatin fr WFS. Its fundamental cnclusin is the fact that a wave field is knwn as sn as the wave frnt n a cntinuus surrunding surface is knwn. The majr step t put this thery int practice is the discretisatin f the surface t a grid f psitins and the reductin f the reprductin dimensins t the hrizntal plane. Accrdingly, the WFS principle states that a wave field in the hrizntal plane can be synthesized when the signals n the psitins f an array f transducers are knwn. An example is depicted in Figure 1: Frm sund surce S (which emits a sine impulse and is lcated in an infinitely large plane withut demarcatin f walls), a wave frnt results as illustrated in Figure 1a). If ne nw places an array f n micrphnes (M) in this primary sund field and ne reprduces the recrded micrphne signals via an equally arranged array f n ludspeakers (L) special equalizatin has t be included accrding t the relevant physical basics in a reprductin rm (Figure 1b), ne btains the synthesized wave frnt in the (red dtted) listening area. At any place in the listening area the listener perceives a virtual sund surce S, as he can mve arund freely; whilst the virtual sund surce remains crrectly lcalized in terms f its directin (see [5] r [8]). a) Listening Area Figure 2: WFS is capable f reprducing bth the stable psitins f pint surces (red and pink, dashed and dtted) and the stable directin f a plane wave (blue, slid) In cntrast t sterephny WFS is able t: - prduce virtual surces that are lcalized n the same psitin thrughut the entire listening area, refer Figure 2: The red (dashed) and pink (dtted) arrws indicate the directins f the auditry events when the red and pink virtual pint surces are reprduced. - prduce plane waves that are lcalized in the same directin thrughut the entire listening area, refer Figure 2: The blue (slid) arrws indicate the directin f the auditry event when the blue plane wave is reprduced. - enhance the lcalizatin f virtual surces and the sense f presence and envelpment thrugh a realistic reprductin f the amplitude distributin f a virtual surce. In ther wrds, when the listener is appraching the lcatin f a virtual surce the amplitude increases in a realistic way. Accrdingly, the amplitude f a plane wave - which can be seen as a surce in infinite distance - changes least n different listener psitins. - prduce fcused surces. Fcused surces that are in principal lcalized in frnt f the array. Figure 3 shws the wave frnts f a pint surce behind the array (a) and in frnt f the array (b) in a simulatin. Naturally, the lcalizatin will nt be crrect fr listening psitins between the fcus pint and the array because the sund emissin f the virtual surce ccurs here reversely. b) Figure 1: Principle f WFS a) ideal surce respnse b) typical utput f a finite WFS array 2.2. Special Prperties f WFS Thrugh WFS the sund engineer has a pwerful tl t design a sund scene. One f the mst imprtant (with respect t cnventinal techniques) nvel prperties is its utstanding capability f prviding a realistic lcalizatin f virtual surces. Typical prblems and cnstraints f a sterephnic image vanish in a WFS sund scene. Figure 3: Wave frnts f virtual surces, after [10] a: behind the array b: in frnt f the array 2

These prperties enable the synthesis f cmplex sund scenes which can be experienced by the listener while mving arund within the listening area. This feature can be made use f deliberately by the sund engineer t realize new spatial sund design ideas. Mrever, it has been shwn that the enhanced reslutin f the lcalizatin cmpared with sterephny [9] enables the listener t easily distinguish between different virtual surces making the sund scene significantly mre transparent. 3. WFS PRACTICAL CONSTRAINTS Nt surprisingly, in practice it is nt pssible t match all theretical requirements fr a perfect result. The rendered WFS sund field differs frm the desired sund field t sme degree fr a number f reasns (fr details see [11]): 3.1. Discreteness f the array (spatial aliasing) Spatial aliasing prduces spatial and spectral errrs f the synthesized sund field due t the discretisatin f the secndary surce distributin. The wave field is prduced errneusly abve the spatial aliasing frequency f alias which depends n the ludspeaker spacing and the surce/listener gemetry. Spatial aliasing gives rise t artefacts f cluratin and lcalizatin. 3.2. Reflectins f the reprductin rm A WFS array can nt render the desired sund field perfectly if reflectins f the reprductin rm prduce interference in spatial perceptin. In particular, perceptin f distance, depth and spatial impressin are affected, because fragile distance cues f synthesised surces can be dminated by the strnger distance cues generated by the array speakers. They interfere with the desired reflectin pattern f the synthesised surce. Special rm cmpensatin algrithms being under investigatin ([12], [13]) will perhaps be able t minimize this effect. 3.3. Restrictin t the hrizntal plane Thery des nt restrict WFS t the hrizntal plane. Hwever, the reductin f the array dimensin t the hrizntal plane is the practical apprach, having a number f cnsequences. First, virtual surces can be synthesized nly within the hrizntal plane. This includes virtual reflectins affecting the cmpleteness f a natural reflectin pattern and thus pssibly resulting in impairments f perceptin f distance, depth, spatial impressin and envelpment. Furthermre, hrizntal arrays d nt generate real spherical waves, but cylindrical waves. In the case f imaging a plane wave fr example there results an errr with respect t the level rll-ff (3dB/dubling f distance), in cmparisn with the ideal plane wave (n rll-ff) ([11], [14]). 3.4. Limitatin f array dimensins (diffractin) In practical applicatins the ludspeaker array will have a finite length. Due t the finiteness f the array, diffractin waves riginate frm the edges f the ludspeaker array ([11], [14]). These cntributins appear as after-eches (and pre-eches respectively fr fcused surces), and depending n their level and time-ffset at the receiver s lcatin may give rise t cluratin. Methds t reduce these truncatin effects are knwn, e.g. by applying a tapering windw t the array signals. This means that a decreasing weight is given t the ludspeakers near the edges f the array. In this way the amunt f diffractin effects can substantially be reduced at the cst f a limitatin f the listening area [14]. 3.5. Effects n perceptin Althugh a number f authrs have suggested methds t deal with the practical limits f rendering accurateness r t minimize their effects, there is still a lack f knwledge (sme details can be fund e.g. in [5], [6], [11], [15]). Several effects f the cnstraints n specific perceptual attributes are nt knwn yet in detail. Hwever, this knwledge is imprtant fr further develpments f WFS systems in view f future applicatins. Current psychacustic studies are cncentrating n the subjective evaluatin f principle characteristics f WFS systems in cmparisn with sterephnic r binaural systems. They are necessary t evaluate the resulting impacts n attributes f spatial perceptin nt nly with respect t the develpment f WFS systems fr different applicatins but als in view f scientific knwledge. Particular attentin shuld be turned t the perceptin f sund clur, directin, distance, spatial depth, spatial perspective, spatial impressin, reverberance, and envelpment. 4. WFS APPLICATIONS 4.1. The Eurpean CARROUSO Prject / 1 The Eurpean CARROUSO Prject ( Creating, Assessing and Rendering in Real Time f High Quality Audi-Visual Envirnments in MPEG-4 Cntext ) intended t break several limitatins f these current cmmercial systems by merging the new WFS rendering technique with the flexible cding standard MPEG-4, allwing bject-riented and interactive sund manipulatin. By means f the MPEG-4 frmat the signal f the surce ( Gestalt ) and its spatial prperties are transmitted separately. Fr reprductin, the dry surce signal is cnvlved with the measured r mdelled set f impulse respnses (cntaining the spatial infrmatin), and emitted by a ludspeaker array. In cntrast t sterephny WFS is able t prduce virtual stable surces lcalized at the same psitin thrughut the entire listening area, prduce virtual surces in frnt f the ludspeaker array ( fcused surces ) prduce plane waves that are lcalized in the same directin thrughut the entire listening area, enhance the sense f depth, spatial impressin and envelpment thrugh a realistic reprductin f the riginal rm respnse The key bjective f the prject CARROUSO was t prvide a new technlgy that enables t transfer a sund field, generated at a certain real r virtual space, t anther usually remte lcated space. The results have shwn the pssibility t capture, transmit and render sund surces and their related acustic envirnment with mre realism, cmpared t existing sterephnic methds. They are cnsidered as a majr milestne fr immersive audi representatin at public places and in private husehlds. Tw applicatins were targeted within this prject. The first ne cncerns high quality spatial audi with assciated vide fr bradcasting. The secnd applicatin is related t cperative and interactive wrk n immersive audi bjects. / 1 EU-Prject IST 1999-20993 (Jan. 2001 June 2003): [4]. 3

4.2. Synthesis f WFS and sterephny Fr recrding f rchestra and slist clsely spaced spt micrphnes are used. The sterephnic rchestra mix shuld be cmpsed in a way that it cntains as little rm infrmatin (reverb, reflectins, etc.) as pssible; but it shuld cntain the adequate spatial distributin f elements. This multi-channel sterephnic mix and the slist signal are cnvlved with the apprpriate spatial impulse respnses (see Figure 4). As a result, the rendered WFS sund field represents stable virtual surces lcated in the cncert hall. Listeners within the listening area perceive a multi-channel sterephnic image f the rchestra and a pint surce image f the slist, whereby the reprduced characteristics f the cncert hall give a new sense f realism. On this basis apparent advantages f established cnventinal sterephnic recrding techniques n the ne hand and f WFS technlgies n the ther can, in principle, can be utilized thrugh a purpseful cmbinatin. Wave field synthesiser (Cnvlutin) Virtual imaging area Orchestra rear imaging plane Orchestra frnt imaging plane Listening area VPS Orchestra Recrding Sterephnic mix Slist Wave field synthesiser (Cnvlutin) Rm respnse measurement Strage Figure 4: WFS: Separate handling f surces and spatial infrmatin 4.3. Virtual Panning Spts (VPS) The key tl is use f s-called Virtual Panning Spts (VPS) [16], virtual pint surces t be applied fr panning acrss any sterephnic imaging plane in the virtual WFS imaging area. VPS can be understd as virtual ludspeakers which reprduce the sterephnic sund image f a spacius sund surce (e.g. a chir) in the recrding rm (see als [17], [18]). The suitable rm impulse respnses have t be measured in the riginal rm r t be created artificially in a suitable way. In the example f Figure 5, the rchestra is imaged with the assistance f six VPS, which are reprduced via WFS and are relatively freely cnfigurable with regard t lcalizatin, expansin and distance. The sund design advantage f this cncept is self-explanatry: The sterephnic recrding f the rchestra accrding t Figure 4 prduces a spacius sund image f the surces as there is an image between the VPS in accrdance with the principles f phantm surces lcalizatin. The ludspeakers are virtual surces, generated thrugh WFS and prvided with the rm characteristics f the recrding rm. The lcatins f the VPS behave directinally stable in the listening area. The knwn disadvantages f phantm surce lcalizatin, especially the lw directinal stability can be easily avided by emplying a sufficient number f VPS. Figure 5: Use f Virtual Panning Spts (VPS) The number f sterephnic imaging areas is in principle arbitrary. Frm an artistic pint f view, ne shuld rientate neself twards the number f spacius instruments r instrument grups (in large ensembles, e.g. string grups, brass player grups, chir). The number and spatial distributin f the VPS depends n the fllwing criteria: Size and shape f the hmgeneus ensemble Circumstances f the prductin Artistic and sund balance-related intentin f the sund engineer Available transmissin capacity Virtual Panning Spts, VPS, are selected pints ( virtual ludspeakers ), which prduce a sterephnic representatin area. These can in principle be selected at chice in accrdance with the recrding situatin and the desired sund image. The imaging area can be spread ut by an arbitrary number f VPS in a randm spatial expansin in accrdance with the situatin and intentin. The artistic arrangement f the ensemble upn the WFS transmissin cmmences with the chice, dedicatin and psitining f the VPS. Three parameters shuld be mentined, which lends the sund engineer t new pssibilities f spatial sund design: 1. In the case that ludspeaker arrays are installed lateral t the listening area, there are in principal, n prblems as far as directinal stability is cncerned, as a lateral sterephnic representatin sectr can be built up frm a sufficient number f stable VPS. The same applies t the sectr behind the listener. 2. A sterephnic imaging area des nt nly allw itself t be mved in all directins, stretched ut r cmpressed, but can als be presented in an extensive range with different distances. The representatin f depth is thus easily recgnizable. 3. With certain cnstraints (see Sectin 4), the VPS can be placed in a distance between the listener and the ludspeaker array and als with the sterephnic imaging field. Thus, the virtual imaging area theretically reaches in dense clseness t the listener and allws fr an expressive representatin f depth. 4

4.4. Virtual ludspeaker reprductin An imprtant applicatin f the VPS technique is a special preset f the VPS setup n the reprductin side, which enables the reprductin f cnventinal multichannel recrdings in a virtual listening rm [16]. Fr this purpse, tw mdificatins are suggested fr the WFS decder, which can be activated in the event f need fr applicatin, see Figure 6: Surrund sund surce (e.g. DVD) Wave field renderer (Cnvlutin) L VPS cnvlutin data: Respnses frm the reprductin rm C R Cnvlutin data: Respnses frm the recrding rm 4.5. Virtual headphne reprductin The standard way t play binaural signals via ludspeakers is t use crsstalk cancellatin (XTC) filters t eliminate the unwanted signal paths between speakers and ears (see e.g. [20]). Ideally, the head related transfer functins (HRTF) f the path between the ludspeakers and the ears f the listener have t be measured r calculated with a mathematical mdel (e.g. [21], [22]), and inverse XTC filtering must ensure that the resulting transfer functin is neutral (= 1). The need fr head-tracking results in the use f a whle set f XTC filters which have t be updated with every head rtatin, see Fig. 7. L Headphne input XTC δ R ( H L1 / δ ; H L2 / δ ; H R1 / δ ; H R2 / δ ) -1 Subw. LS Listening area RS H L2 / δ H L1 / δ H R2 / δ! = 1 H R1 / δ Figure 6: VPS cnfiguratin fr rendering virtual 5.1 multi-channel ludspeakers Figure 7: Headtracking cntrlled inverse HRTF filtering (XTC δ ) t prvide a neutral transfer functin f the virtual headphne 1. The cnfiguratin f the VPS with regard t rm impulse respnses and spatial arrangement is dne in accrdance with the preset setup f virtual ludspeakers in a virtual listening rm. Arbitrary arrangements f the virtual ludspeakers can be preset and be activated dependent n the sterephnic frmat t be reprduced. 2. The virtual surce signal is nt received via the transmissin channel, but frm the multichannel decder n the reprductin side (e.g. that f a DVD player). The WPS reprductin unit perates cmpletely detached frm WFS transmissin, and can principally ffer three attractive advantages: 1. Diverse sterephnic multichannel frmats can be easily reprduced ptimal thrugh the selectin f a VPS preset, withut having t apprpriately adjust the ludspeaker arrangement within the living-rm. 2. The virtual ludspeakers can als be placed utside the living-rm, i.e. als in a cnfined area situatin, the listening area fr multi-channel sterephny is sufficiently large. 3. A future high quality WFS reprductin unit will allw fr an electrnic cmpensatin f diverse defects in the reprductin rm [12], especially the reductin f the effect f the early reflectins and the balancing f asymmetrical arrangements f the speaker array. Frm the technical and practical pint f view the applicatin f WFS fr multichannel stere reprductin culd be the first step twards acceptance in the market place. In this regard, the develpment f the s-called MAP technlgy (see [19]) is imprtant. The flat panels, e.g. fed with glass fibre cables, can ften be better integrated int the living-rm and are mre attractive than cnventinal ludspeakers. 4.5.1. Cncept The realizatin f a headtracking cntrlled inverse HRTF filtering is prblematic particularly in case f larger head rtatins. In the best case, the ludspeakers ught t mve with the head rtatin, s that the relative psitins between the sund surces and the ears stay cnstant and nly ne set f XTC filters is needed. Of curse this cncept is nt feasible with real ludspeakers. L Headphne input XTC 0 WFS fcused surce rendering Figure 8: Headtracking cntrlled WFS rendering f fcused surces t prvide a neutral transfer functin f the virtual headphne withut the need f adaptive XTC filters R ( H L1 / 0 ; H L2 / 0 ; H R1 / 0 ; H R2 / 0 ) -1 H L2 / 0 H L1 / 0 H R2 / 0 H R1 / 0 = 1 5

WFS hwever enables t replace the real speakers by fcused surces that act as the transaural ludspeakers and can easily be mved arund by adjusting the driving functins (i.e. the delay times and attenuatins) f the array ludspeakers. By synthesizing fcused surces at a clse distance t the listener's head a stable virtual headphne reprductin can be achieved withut the need f adaptive XTC filters (Figure 8). The actual cnfiguratin f the fcused surces is adaptable fr ptimum inverse filtering perfrmance. Instead f a standard linear WFS array hwever, a circular design ensures cnstant distances between the ears, the fcused surces, and the array speakers. This leads t a cnstant (and high) aliasing frequency and greatly reduces audible sund clratins during head rtatins. Figure 9 (left) shws a schematic view f the circular array rendering tw fcused surces (red and blue dts). The dashed circle indicates the path n which the fcused surces mve during a full head rtatin. Figure 9: The circular array lcated abve the listener synthesizes fcused surces which act as transaural surces. The array setup is suspended abve listener s head and includes a lw frequency driver in the middle f the circle (Figure 9, right). N ludspeakers in the listener's field f visin are bstructing the view n e.g. cmputer displays r TV mnitrs. 4.5.2. The Binaural Sky An attractive applicatin f the virtual headphne is reprductin f binaural signals, particularly reprductin f Binaural Rm Synthesis signals (BRS, see e.g. [23], [24], [25]). The BRS headphne playback system was develped in the 1990s by the Institut für Rundfunktechnik IRT. Cntrlled by headtracking data, multi-channel audi signals are cnvlved with measured r mdelled binaural impulse respnses f a mnitring rm. The resulting headphne input signals ffer accurate rm-related lcalizatin f the virtual ludspeakers in terms f directin and distance regardless f listener s head rientatin. The BRS-system is a perfect tl fr accurate surrund sund headphne reprductin. Irrespectively f the actual acustic prperties f an individual cntrl rm, like in a small OB-van, the sund engineer can wrk in a virtual acustic envirnment which is identical t his familiar r preferred envirnment. Cnsumers can use a (virtual) high-quality studi fr surrund sund reprductin. Hwever, a drawback f the BRS system is the need t wear headphnes. Fr a number f applicatins and situatins, e.g. in a car r in a situatin where real and virtual surces have t be mixed ( augmented reality ), a binaural reprductin withut headphnes wuld be desirable. Fig. 10: The Binaural Sky cncept BRS reprductin via WFS based virtual headphne BRS reprductin withut wearing headphnes is the idea and intentin f the nvel system knwn as Binaural Sky [26] which is the cmbinatin f the tw headtracking based cncepts, BRS and virtual headphne as intrduced in chapter 4.5.1. As indicated in Fig. 10, the WFS array abve the listener generates fcused surces clse at the ears. They are applied as transaural surces frming a virtual headphne which is used fr BRS reprductin. As a result, stable and accurate rm-related lcalizatin f virtual ludspeakers in the hrizntal plane is achieved. The lcatin f the array ludspeakers is nt perceivable. Results f subjective and bjective tests n a first realizatin f the Binaural Sky setup have been reprted in [26]. In the upper hemisphere (abve hrizntal plane) they shw gd lcalisatin f virtual surces in terms f directin and distance, cmparable t BRS playback via real headphnes r even real sund surces. Hwever, similar t experiences with dummy head technique, surces in the hrizntal plane are frequently heard abut 10 t high, and belw the hrizntal plane perceptin f depth is prblematic. Variatins in sund clur between different virtual surces are uncritical. BRS reprductin can be enjyed withut wearing a headphne and withut any ludspeaker in the listener's field f visin. Thus the system is perfectly suitable fr applicatins in the field f virtual reality. 5. REFERENCES Virtual surce [1] Theile, G.: On the lcalisatin in the superimpsed sundfield. PhD Thesis, Techn. University Berlin, 1980. nline www.hauptmikrfn.de/theile/theile_thesis.pdf [2] Theile, G.: "On the Naturalness f Tw-Channel Stere Sund". J. Audi Eng. Sc. 39, 1991, pp. 761-767 [3] Theile, G.: "Multichannel natural music recrding based n psychacustic principles". AES 19th Intern. Cnf. 2001, Prceedings pp. 201 229, updated versin see www.hauptmikrfn.de/theile/mcrecrding.pdf [4] Brix, S.; et al: CARROUSO An Eurpean Apprach t 3D- Audi.110th AES Cnv., 2001, Preprint 5314 [5] Berkhut, A. J.; de Vries, D.; Vgel, P: Acustic Cntrl by Wave Field Synthesis, J. Acust. Sc. Am., Vl. 93, pp. 2764 2778, 1993. [6] Bne, M.M, Verheijen, E.N.G.; van Tl, P.F.: Spatial Sund Field Reprductin by Wave Field Synthesis. J. Audi Eng. Sc., Vl. 43, pp. 1003-1012, 1995 [7] Geiger, H.; Scheel, K.: Handbuch der Physik, Band VIII. Verlag Julius Springer, Berlin, 1926 [8] Hrbach, U.; Bne, M.: Practical Implementatin f Databased Wave Field Reprductin System. 108th AES Cnv., 2000, Preprint 5089 6

[9] Bne, M.; de Bruijn, W.: Imprving Speech Intelligibility in Telecnferencing by using Wave Field Synthesis, 114th AES Cnv., 2003, Preprint 5800 [10] Bne, M.: Acustic Rendering with wave field synthesis, ACM SIGGRAPH and EUROGRAPHICS campfire, Snwbird, Utah, May 26-29, 2001 [11] Wittek, H., 2003: Perceptin f Spatially synthesized sund fields Literature review abut WFS, available nline at: www.hauptmikrfn.de/wfs.htm [12] Sprs, S.; Kuntz, A.; Rabenstein, R.: An apprach t listening rm cmpensatin with wave field synthesis. AES 24 th Intern. Cnf., 2003. Prc. pp. 70 82 [13] Crteel E., Nicl R., Listening rm cmpensatin fr Wave Field Synthesis. What can be dne? AES 23rd Int. Cnf., Cpenhagen, 2003 [14] Snke, J.-J.: "Variable Acustics by wave field synthesis", Thela Thesis, Netherlands, 2000, ISBN 90-9014138-3 [15] Verheijen, E.: Sund Reprductin by Wave Field Synthesis. Delft University f Technlgy, 1997 [16] Theile, G.; Wittek, H.; Reisinger, M.: Ptential Wavefield Synthesis Applicatins in the Multichannel Sterephnic Wrld. AES 24 th Intern. Cnf., 2003. Prceedings pp. 43 57 [17] Bne, M.; et. al.: Virtual surrund speakers with wave field synthesis. 108th AES Cnv., 1999, Preprint 4928 [18] Kuhn, C.; Pellegrini, R.; Leckschat, D.; Crteel, E.: An Apprach t Miking and Mixing f Music Ensembles Using Wave Field Synthesis. 115th AES Cnv., 2003 [19] Bne, M.M.: Multi-Actuatr Panels (MAPs) as ludspeaker arrays fr wave field synthesis : J. Audi Eng. Sc, 52 (7-8), pp. 712-723. [20] Bauck, J.; Cper, D. H.: Generalized Transaural Stere and Applicatins. In: J. Audi Eng. Sc. 44 (1996), pp. 683 705 [21] Damaske, P: Head related tw channel sterephny with ludspeaker reprductin. JASA 50, 1109-1115, 1971 [22] Gardner, W. G.: 3-D Audi Using Ludspeakers. PhD Thesis, Massachusetts Institute f Technlgy, 1997 [23] Hrbach, U.; Pellegrini, R.; Felderhf, U.; Theile, G.: Ein virtueller Surrund Sund Abhörraum im Ü-Wagen. Prceedings 20. Tnmeistertagung, 1998, ISBN 3-598-20361-6, pp.238-245 [24] Mackensen, P.; Fruhmann, M.; Thanner, M.; Theile, G.; Hrbach, U.; Karamustafaglu, A.: Head-Tracker Based Auralizatin Systems: Additinal Cnsideratin f Vertical Head Mvements. AES 18th, 2000, Preprint 5135 [25] Spikfski, G.; Fruhmann, M.: Optimisatin f Binaural Rm Scanning (BRS): Cnsidering inter-individual HRTFcharacteristics. In: AES 19th Int. Cnference n Surrund Sund, Elmau, 2001 [26] Menzel, D.; Wittek, H.; Theile, G.; Fastl, H.: The Binaural Sky: A Virtual Headphne fr Binaural Rm Synthesis. Tnmeistersympsium, Nv. 2005, Paper available nline at www.tnmeister.de/sympsium/2005/np_pdf/r04.pdf 7