Objective measures of spatial sffects in Spanish concert halls

ARCHIVES OF ACOUSTICS Vol. 37, No. 4, pp. 529–547 (2012) Copyright c 2012 by PAN – IPPT DOI: 10.2478/v10168-012-0063-y Objective Measures of Spatial Effects in Spanish Concert Halls Sara GIRÓN(1), Teófilo ZAMARRE˜ NO(1), Pedro BUSTAMANTE(2) (1) Departamento de F´ısica Aplicada II Universidad de Sevilla, E.T.S. Arquitectura (IUACC) Av. Reina Mercedes 2, 41012-Sevilla, Spain; e-mail: {sgiron, teofilo}@us.es (2) Departamento de Construcciones Arquitectónicas I, Universidad de Sevilla, E.T.S. Arquitectura (IUACC) Av. Reina Mercedes 2, 41012-Sevilla, Spain; e-mail: bustaman[email protected] (received April 9, 2012; accepted October 22, 2012) The present work consists of a statistical study of the monaural (lateral-reflection fractions and level) and binaural acoustic parameters (inter-aural cross-correlation coefficients) that evaluate the amount of early and late lateral acoustic energy encountered in 9 performance halls in Andalusia (southern Spain). Hall volumes range between 6,163 m3and 34,594 m3and all enclosures are used for presentations of symphonic concerts and other music performances. The majority of these venues are located in provincial capitals of the community and often constitute the only premises in the city where symphonic concerts can be held. The acoustic parameters under study here were derived from impulse responses analyses using a sine-sweep signals which were generated and processed by WinMLS 2004 software in the octaveband frequency centred from 125 to 4 kHz, and all parameters were spectrally averaged according to the ISO 3382-1 standard. A comparison is presented of monaural experimental results as a function of sourcereceiver distance with the prediction of Barron’s revised theory for concert halls, and the analyses of the acoustic parameter results are carried out in terms of their respective just noticeable differences: at the many microphone positions for the two source positions on stage, for the spatial distribution of seats in the audience zone relative to the central axis (for leftand right-hand sides) of the rooms, and for the presence of the orchestra shell on stage. Results reveal that the orchestra shell propitiates a perceptible decrement in the values of the early lateral energy fraction and an increment in the late lateral level at the audience seats. In addition, a regression study reveals that the two kinds of measures of laterality, monaural and binaural, are correlated when the hall-average data is considered, but they remain uncorrelated when all individual positions are used. Likewise, the ranges of variation of the acoustic parameters found in these halls are narrower than those specified in the ISO 3382-1. The paper concludes with a discussion on the relationships of hall-average data of the five parameters with eight geometric and acoustic variables. Keywords: concert hall acoustics, directional energy parameters, binaural parameters, spatial impression, spatial effects. 1. Introduction After the commencement of the science of architectural acoustics with W. C. Sabine’s contributions in 1895 (Sabine, 1993), breakthroughs and rapid development of electroacoustics in the first half of the twentieth century propitiated a great advance in general acoustics. However, it is from the last third of the last century that the science of room acoustics experienced a resurgence propelled by the emergent computer and signal science. As a consequence of research in that new age, considerations other than those in the monaural temporal factors came into play beyond the scope of reverberation time to determine optimal configurations for listening spaces (Beranek, 1962). In the 1960s, the first references appeared (Schroeder et al., 1966; Marshall, 1967) that considered the significance of spatial effects in the acoustics of concert halls. This perception refers to how the acoustical impression in a room differs from that which lacks many delayed reflections arriving from all directions. Soon after, work by Marshall (1968), and Barron (1971) confirmed that this impression is produced by the early lateral reflections, and this understanding has since had a significant impact on concert hall design (Cremer, 1989; Beranek, 1992). As a typical spatial factor of the sound field, Damaske and Ando (1972) defined the interaural cross-correlation coefficient, and in a refined 530 Archives of Acoustics – Volume 37, Number 4, 2012 approach, Ando (2007) developed a theory of subjective preferences in relation to temporal and spatial factors of the sound fields based on the modelling of the human auditory-brain system (Soeta et al., 2002). Rooms for music are the most visible and interesting spaces in architectural acoustics where the three disciplines, of acoustics, architecture and music are blended. However, rating the acoustic quality of a hall for musical performances is a non trivial and multidimensional topic due to the lack of an unequivocal criterion of listeners’ music appreciation, in contrast to that for the one-dimensional comprehension of speech (Peutz, 1971). Spatial impression phenomenon is a general term which covers two subjective perceptual categories: apparent source width (ASW), and listener envelopment (LEV). ASW is the apparent auditory width of the sound field created by a performing entity as perceived by a listener in the audience area of a concert hall, and experiments have concluded that it is mainly related to the early lateral reflections reaching the listener (Barron, 1971). LEV is considered as the subjective impression by a listener of being enveloped by the sound field: a condition that is, according to experiments, primarily related to the late lateral sound field (Bradley et al., 2000). Two types of measurements of lateral reflections from the impulse response in a room have emerged: the monaural lateral energy fractions (J) related to the energy of early or late lateral reflections in the former case, and the interaural cross-correlation coefficient (IACC) in the latter, which is a measure of the difference in sound at the two ears and hence of laterality. To assess ASW, the early lateral energy fraction JLF (Barron, Marshall, 1981), (or alternatively, the early lateral energy fraction cosine JLFC (Kleiner, 1989), is accepted as a monaural directionally influenced measure. Likewise, the normalised interaural cross-correlation function IACF (Damaske, Ando, 1972) is a measure of the dissimilarity between the arrivals of the wave at the two ears. From this function the interaural cross-correlation coefficients, IACC, are then given by: IACCt1, t2= max IACFt1, t2for −1ms < τ < +1 ms. In which t1= 0 and t2= 80 ms (subscript E) are chosen for the early interaural cross-correlation coefficient IACCEwhich assesses the ASW perceptual attribute (Hidaka et al., 1995; Okano et al., 1998). In the same way, the late lateral sound level LJis proposed as a standard (Bradley, Soulodre, 1995a; 1995b) in the monaural impulse response measurement in order to provide a practical indicator of the amount of listener envelopment in concert halls for sound arriving from all directions (Evjen et al., 2001), and the late interaural cross-correlation coefficient IACCL (correlation calculated from t1= 80 ms to infinity (subscript L)) in the binaural measurement, is proposed by Hidaka et al. (1995), and Okano et al., (1998). Alternatively, a further parameter which involves a ratio, the late lateral energy fraction JLLF, (Barron, 2001) defined by the same ratio as JLF but for the late interval, to study listener involvement is also considered here. Both monaural late lateral parameters are related to omnidirectional sound strength Gand clarity parameter C80 through: JLLF = 10(LJ−G)/10 1 + 10C80/10.(1) Due to the contamination of the results with background noise under certain frequencies, this parameter cannot be obtained directly from WinMLS, and hence its calculation in this work is as defined above. According to several experiments, the spectral content of lateral reflections that contribute towards the ASW and LEV and lateral levels in the monaural parameters are in the four octave bands with centre frequencies 125, 250, 500, and 1000 Hz (Hidaka et al., 1995), respectively. This measure favours the frequency region where the wavelengths are mostly longer than the acoustical distance between the leftand righthand sides of a head. In the case of the IACC parameter, no widespread agreement yet exists on how to calculate averages over various octave bands: in fact ISO 3382-1:2009(E) (2009) does not specify this point. Throughout this work the spectral average of the IACC parameter is determined following Hidaka’s instructions (Hidaka et al., 1995; Okano et al., 1998), which are based on previous experiments of Blauert and Lindemann (1986a, 1986b) and which concludes that 500, 1 k, and 2 kHz are the octave bands of equal and principal importance for the ASW attribute covering the effect on spatial impression categories of the entire octave band frequency range (Okano et al., 1998). Henceforward, these experimental measures obtained by arithmetically averaging within these respective aforementioned frequencies are named JLFm, JLFCm,JLLFm, IACCEm, and IACCLm, respectively. According to international standards (ISO, 2009) the frequency-averaged late lateral sound energy level, LJm must be averaged in energy. Although De Vries et al. (2001) suggested a decomposition of the measures to suppress interference effects to which the human ear is apparently insensitive, the revised ISO standard (ISO, 2009) has proposed JLFm,JLFCm as objective measurements of the perceptual listener aspect ASW, a just-noticeable difference of 0.05, and a typical range of values from 0.05 up to 0.35 (frequency-averaged values in single positions in non-occupied concert and multi-purpose halls up to 25,000 m3). For LJm, its just-noticeable difference remains unknown and for IACC according to Cox et al. (1993) a just-noticeable difference of 0.075 is assumed. S. Girón, T. Zamarre˜no, P. Bustamante – Objective Measures of Spatial Effects in Spanish Concert Halls 531 Within this field, other pieces of work (Morimoto et al., 2008) have proposed experiments to clarify the essentials of ASW and LEV from the point of view of auditory behaviour by suggesting that the acoustic components of the reflections over and beyond the upper limit of the precedence effect contribute towards ASW and LEV, respectively. Furthermore, the components from behind the listener generate greater LEV (Morimoto et al., 2001). Abdou and Guy (1996) also pointed out the necessity of spatial information on the sound fields for the evaluation of the acoustics of rooms and the diagnosis and correction of the causes of acoustic defects. This paper aims to interpret and discuss the results of monaural and binaural measures of spatial impression in 9 performance spaces of southern Spain which are used as concert halls and for other musical performances. Analyses are carried out in terms of seat positions and hall-average values with significant physical variables of the acoustic field. The spatial distribution of the various parameters, the differences found due to the two source positions on the stage, and to the orchestra shell are analysed in terms of their respective just-noticeable differences (JND). These analyses could provide a starting point for further studies of correlation between objective parameters and spatial impression. 2. Experimental method 2.1. Measurement technique The procedures employed here are those established in the ISO 3382-1 standard (ISO, 2009), whereby measurements were carried out in unoccupied rooms and on unoccupied stages (no public, no musicians), safety curtains were always drawn open, and, in certain cases with the orchestra shell configuration in the stage, chairs and music stands were present. Temperature and relative humidity were monitored during the measurements which revealed a range of variation of the environmental conditions during measurements in all enclosures of 16.8–25.2◦C for the temperature, and 38–60% for the relative humidity, respectively, with the exception of Falla Grand Theatre which was measured in winter, whose minimum temperature was 14.1◦C. The variations between the environmental conditions for the halls that have been measured with and without an orchestra shell in the stages never exceeded 1◦C in temperature and 8.5% in relative humidity respectively, except for Falla Grand Theatre whose measurements were carried out in winter and spring with a difference of 9◦C in temperature and 12% in relative humidity, respectively. Monaural and binaural impulse responses (IRs) were measured to determine, among others, the following parameters for each frequency band between 125 Hz to 4000 Hz and in all receiver positions: the early lateral energy fraction (JLF )and early lateral energy fraction cosine (JLFC), the late lateral fraction (JLLF), the late lateral level (LJ), and the binaural interaural cross-correlation coefficients (IACC), to study the spatial impression phenomena in the rooms. The IR has been obtained at each reception point using sine sweep signals which were generated and analysed by WinMLS 2004 software via the VX Pocket v2 sound card from Digigram. The omnidirectional dodecahedral source DO12 with its InterM-1000 amplifier, is placed at the most usual point of location of the natural source (S) on the orchestra platform (none in the orchestra pit which was always covered with wooden panels or heavy curtains) at a height of 1.50 m from the floor. Wherever possible, measurements in two source positions on the stage have been carried out (see Table 1 and Fig. 1). In all cases the S1 position coincides with the central source position on the stage; and these two source locations coincide for the configurations with and without an orchestra shell. Two microphones have been used: multipattern configurable Audio-Technica AT4050/CM5 (with an appropriate signal-conditioning Earthworks LAB1 amplifier) to measure lateral energy fractions and levels; and a torso type HSU III simulator (Code 1323) from Head Acoustic (with OPUS amplifier) to record binaural IR. They are all located at the approximate height of the head of a seated person, ∼1.20 m from the floor, in a predetermined number of seats within the various audience zones (these seats are coincident for the measurements carried out with and without the orchestra shell on the stage in certain halls). 2.2. The halls researched In this section, data for the 9 concert halls under research is presented: Cordoba Grand Theatre (CGT), located in Cordoba; Falla Grand Theatre (FGT), located in Cadiz; Huelva Grand Theatre (HGT), located in Huelva; Isabel la Católica Theatre (ICT), located in Granada; Lope de Vega Theatre (LVT), located in Seville; Miguel de Cervantes Theatre (MCT), located in Malaga; Manuel de Falla Auditorium (MFA), located in Granada; Maestranza Theatre (MT), located in Seville; and Villamarta Theatre (VT), located in Jerez de la Frontera (Cadiz). They appear in alphabetical order in accordance with the acronyms assigned from their names on the MIREM web site (digital scenic and musical enclosure map of Spain), where further information on the enclosures is available (MIREM, 2012). Details on the stage shell of these spaces and stage support parameters have been published elsewhere (Girón et al., 2010). In all these buildings, several types of cultural activities, such as opera, theatre, dancing, cinema and concerts, may be performed, except in the Manuel de Falla Auditorium where only concerts and recitals are presented. These are performance spaces with a fixed 532 Archives of Acoustics – Volume 37, Number 4, 2012 public-stage relationship and a frontal description, except the Manuel de Falla Auditorium which presents a bi-frontal description, (it consists of two rooms A and B, see Fig. 1). It is also pertinent to add that all 7 proscenium theatres were refurbished by the regional government during the 1980s and 1990s, and that acoustic measures have been carried out after these refurbishments. In Table 1, the year of completion or inauguration, architectural type, the number of floors, the number of points of reception for the microphones, and the seat capacity are all summarized in alphabetical order, together with the most relevant architectural and acoustical dimensions (León et al., 2007), whose calculations are described below. Mean width W: for horseshoe-shaped halls this is the maximum width on the ground floor excluding boxes; if it is rectangular but irregular, then a mean value is taken as in ICT; in the case of MFA, this geoFig. 1. Ground plan and longitudinal section of 4 halls of the total spaces analysed. The positions of the source and the reception points on each floor are also shown. metrical factor excludes boxes at both sides of room A; and in MT this is its transverse diameter. Mean height H: calculated as the average of the height of the room in the mouth of the stage, in the middle, and at the rear of the hall. Mean depth D: the average of the horizontal measure that exists from the stage to the back of the room on all floors. (In the case of MFA, it is the length of room A). Audience surface SA: the area occupied by all seats in the hall. Total volume V: hall and stage volume (in the two configurations if it exists with and without the orchestra shell). Reverberation time T: the reverberation time measured in the unoccupied halls and averaged spatially and spectrally in the 500 and 1000 Hz octave bands. Absorption A: the absorption obtained using the reverberation time in Sabine’s formula. S. Girón, T. Zamarre˜no, P. Bustamante – Objective Measures of Spatial Effects in Spanish Concert Halls 533 Table 1. Relevant data of the concert halls studied. Hall CGT FGT HGT ICT LVT MCT MFA MT VT Year of completion 1873 1910 1923 1952 1929 1870 1978 1991 1928 TypologyaHP HP SP SP HP HP RA CA SP Number of seats N946 1038 601 689 733 1058 A: 890 B: 413 1800 1200 Hall volume [m3] 6071 8114 4800 5035 5902 6594 A: 4274 B: 1536 20321 7988 Number of floors 4 4 4 3 4 4 1 1 3 Stage volume no shell [m3] 4631 5556 1963 1742 3363 4907 3421 14273 3703 Stage volume with shell [m3] 872 1728 – – – 945 – 1850 574+405 Total volumeb[m3] no shell V′10702 13670 6763 6777 9265 13873 9231c34594 11691 Total volumeb[m3] with shell V6943 9842 – – – 9911 – 22171 8562 Hall mean width, W[m] 17.0 17.7 10.4 18.0 16.6 17.7 A+B: 19.7 37.0 22.8 Hall max depth, D[m] 25.3 25.6 17.4 21.9 27.1 23.3 A+B: 24.4 36.3 24.9 Hall mean height, H[m] 12.4 14.9 16.7 17.3 18.9 19.9 A+B: 11.1 14.3 14.9 Seat surface, SA[m2] 536 687 354 449 411 699 A+B: 786 1156 555 R. points no shell 22 15 16 12/12 14/14 15 22/22 – 18/18 with shell 17 15/15 – – – 15/15 – 15/15 17 T[s] no shell 1.17 1.89 1.41 1.26 1.44 1.26 2.33 – 1.85 with shell 1.19 1.86 – – – 1.14 – 2.51 1.70 A[m2]no shell 1473 1165 772 866 1036 1773 638 – 1017 with shell 939 852 – – – 1400 – 1422 811 aHP – Horseshoe proscenium floor plan. SP – Shoebox proscenium floor plan. RA – Rectangular auditorium floor plan. CA – Cylindrical auditorium floor plan. bHall and stage volumes. cHalls A and B plus stage. To complete this information and to avoid excessive length, Fig. 1 shows the plotted graphs of the ground plans and longitudinal section of only the CGT, MFA, MT, and VT, respectively. The plans include the points of reception on each floor in the theatres with different Fig. 2. Views from the stage of the interiors of the halls; for MFA the view is from room B. symbols, and the two positions of the source on the stage. In addition, Fig. 2 shows a set of photographs of the interiors of each hall taken from their stage, and from room B in the case of MFA. 534 Archives of Acoustics – Volume 37, Number 4, 2012 3. Analyses of experimental results and discussions This section covers, in its various subsections, the analyses of monaural and binaural parameters studied in the 9 concert venues, both in terms of their spatial values in the spaces as well as in terms of the relationships between the different types of parameters that evaluate the same perceptual attributes of spatial impression. The correlations between the hallaverage parameters with the geometric and acoustic parameters of the halls are also carried out. In all cases, the analyses presented refer to the results of the spectral averages of the parameters as stated at the end of Sec. 1. These results can be used to further studies of correlation between objective parameters and spatial impression. As a starting point, a comparison by regression of the experimental values of the two monaural early lateral fractions, JLFm and JLFCm, is carried out. Regression analysis shows that the two parameters are linearly related when the results are compared for each hall and include the results of the two stage configurations, and when the results are studied in all individual positions in all the halls together. In the former case the weakest correlation is, for FGT without the orchestra shell: JLFCm = 0.569 JLFm + 0.132, R2= 0.6144, P = 0.0005, (2) and the strongest linear correlation also for FGT, but with the shell: JLFCm = 0.785 JLFm + 0.104, R2= 0.9560, P < 0.0001. (3) By considering the results in all individual positions for all halls (321 pairs of values, for the two configurations of the stage and the two source positions) the best fit corresponds to a straight line: JLFCm = 0.837JLFm + 0.091, R2= 0.8554, P < 0.0001. (4) The good results of statistical analysis indicate that these two parameters are very similar and that the experimental procedure is correct and can therefore be used alternately for either directional parameter of early lateral energy. Similar results have been obtained with different equipment in worship spaces of a common typology (Girón et al., 2008). Henceforward, the results of JLFm are used since this parameter presents the best facilities for measuring and has experienced the widest use in other experimental work, thereby facilitating comparison. 3.1. Spatial distribution of the lateral acoustic parameters. Effect of the orchestra shell As a first step of this analysis, all the parameters were studied as a function of source-receiver distance. In general, the empirical data set appears as a cloud of points and fails to indicate any predictable behaviour in terms of source-receiver distance. Furthermore, for the two possible stage configurations, these experimental results were also compared with the theoretical values determined by Barron’s revised theory of sound propagation in halls (Barron, Lee, 1988), where reverberation time T, (measured in the unoccupied halls and averaged spatially and spectrally in the 500 and 1000 Hz octave bands), and the total volume V, are the only magnitudes involved in their calculations. In order to avoid excessive length, the detailed results of this study are not shown. However, its main conclusions can be summarized as: A) There is considerable scatter from the theoretical predictions and the furthest deviations generally occur in the late lateral level LJm regardless of the type of hall and of the source-receiver distance. B) The influence of the orchestral shell on the stage in the theoretical predictions is very small. In the experimental results the differences are small, similar to the findings obtained by Bradley (1996), but perceptible (in terms of JND), as will be analysed later. C) Barron’s revised theory of sound propagation in halls indicates that late lateral sound levels are more sensitive to changes in reverberation time and room volume than early lateral fractions (Barron, 2001), and hence the effect of adding an orchestra shell may be the greatest on the late lateral arriving sound levels at audience seat locations. In the absence of predictable behaviour when studying the results of these five parameters versus distance to the source on the stage, it was decided to conduct a study of their spatial distribution by following a common methodology for the five parameters studied. This consists of evaluating the parameters according to distances from the points of reception to the central axis of the room, x(for leftand right-hand sides), and by normalizing these distances with respect to half the average width of each room, w, then these relative distances are presented as (x/w). This method of evaluation is justified by the fact that spatial impression is related to the presence of early and late lateral reflections, and that the closer the walls, the higher intensity should be. For the analysis of this comprehensive spatial distribution in all rooms, the values of the acoustic parameters related to spatial impression are grouped into discrete intervals of the aforementioned relative distances (x/w). The magnitude of these intervals is set at 0.25, where positions x/w > 1express the positions of the microphone very close to the side walls, which in all cases correspond to positions in areas of boxes and S. Girón, T. Zamarre˜no, P. Bustamante – Objective Measures of Spatial Effects in Spanish Concert Halls 535 side terraces on the ground or upper floors and therefore usually under an overhang, (e.g. in CGT, FGT, HGT, LVT, MCT and MFA). It should be borne in mind that all the parameters studied at each position have been averaged in frequency at the octave bands, with the procedures established by ISO 3382-1 (ISO, 2009), listed in Sec. 1. On the other hand, in small cities, it is very common to adapt performance halls to the needs of orchestral music by means of orchestra shells. Despite their widespread use, very little published information exists on their acoustical properties and their influence in the acoustic field of the hall. This section also analyses the influence of shells on the parameters considered in the different areas. 3.1.1. Early lateral energy fraction JLFm In Fig. 3a, the early lateral energy monaural parameter JLFm is plotted in ordinates and scaled at intervals of 1 JND. In abscissa, the relative distance to the central axis of the room is given for all the halls studied without the orchestra shell and with results for the two positions of the sound source (S1 and S2 on the stage, in the drawings of Fig. 1). The number of measurement points in each discrete positional zone is also specified at the top of the graph. The typical range for this parameter (expressed in Fig. 3a as ISO) according to ISO 3382-1 (ISO, 2009), is (0.05, 0.35) for non-occupied concert halls and multi-purpose halls with a volume of less than 25,000 m3. Figure 3a also indicates another narrower range found in these performance spaces (0.10, 0.30) which means a cut of 1 JND in both the upper and lower limits of the range with respect to those specified in ISO 3382-1. It can be observed that this range includes, in each case, the vast majority of experimental results. The criterion used in their specific determination according to Fig. 3b, is discussed later. The filled symbols linked by lines show the average values of the JLFm parameter in each positional interval, and the degree of spatial dispersion in each interval is characterized by the standard deviation (vertical bars). The spatial dispersion tends to increase slightly for positions closer to the side walls x/w >1 and the analysis of the results indicates that, for x/w between 0.25 and 0.75, the average values of the parameter are very constant and only for the area near the centre of the halls, x/w < 0.25, does JLFm reduce its values by 0.5 JND. As a complement to Fig. 3a, and in order to normalize the experimental data with respect to the spatial distribution in the areas concerned, Fig. 3b expresses in ordinates the percentage of the experimental data in each interval of 1 JND (n) for values of JLFm in each positional area (x/w) in relation to the total data in the same area, named n∆(x/w)(see top of Fig. 3a). a) b) Fig. 3. a) Early lateral energy fractions in the relative distance intervals to the central axis of the halls, for all halls, for the two source positions and for the without shell on the stage. The mean values and their standard deviations for each interval are also shown. The typical range of values in this work and in ISO 3382-1 are presented; b) fraction of data for values of JLFm in steps of 1 JND in the different intervals of x/w relative to the total number of data in each interval corresponding to the results shown in part (a). It can be seen that for each interval the total sum fails to reach 100%, the reason being that only those values of the acoustic parameter for which the data lies mostly above 10% in the whole range variation were considered as significant in the two figures. Residual values correspond to the range of variation of JLFm between (0.00, 0.10) in the lower range and (0.30, 0.45) in the top range. This determines the typical range of parameter variation in these rooms. In this distribution, it can be observed how the percentage for the two major ranges (black and dark grey bars) increments when approaching the lateral walls. In order to discern the effect of the orchestral shell on the behaviour of the parameter, Fig. 4a shows a similar study of JLFm but considering only the four performance halls (CGT, FGT, MCT, and VT) which 536 Archives of Acoustics – Volume 37, Number 4, 2012 a) b) c) Fig. 4. a) Early lateral energy fractions in terms of the different intervals of relative distance to the central axis for 4 halls, at S1 source position and for the two stage configurations. The mean values and their standard deviations for each interval and the typical range of values in this work (with shell and no shell) and in ISO 3382-1, are also shown; b) fraction of data for values of JLFm in steps of 1 JND in different intervals of x/w relative to the total amount of data in the same interval, for 4 halls without the orchestra shell; c) idem with the orchestra shell on the stage. have been studied with the two configurations of the stage and solely at the source position S1, to ensure compatible data in the two configurations. The study thus shows the possible effect when the amendment of only one variable is involved (the shell). (Although the amount of data should coincide, there remains a slight difference due to technical problems that arose in one space (CGT), see the top of Fig. 4a). Results confirm the trend obtained for the withoutshell configurations that were studied for all halls in Fig. 3a, and the maintenance of the range established (0.10, 0.30) for both statistical populations of data (thereby validating statistical results for the with-shell configuration), and also show that shells lead to a significant decrease in the average values of the parameter JLFm in all audience zones. Specifically, in the centre of the stalls, x/w ∈(0.00, 0.25), while between 0.50 and 0.75, the decrease is about 0.5 JND. For the zone of boxes and upper balconies under overhangs, the order of the decrease is 1 JND. This suggests, at least initially, that shells focus sound by decreasing the initial lateral reflections, especially in the areas under balconies. In addition, spatial dispersion presents similar behaviour remaining very similar in the other areas and reaches its highest value, at x/w >1, the area near the side walls. Results also confirm the narrower range deduced before for the no-shell configuration in this parameter (0.10, 0.30), and another even narrower range in the upper limit (0.10, 0.25) for the with-shell configuration. The effect of the presence of the orchestra shell on this early lateral energy parameter in the audience seats although small is audible, as seen in Bradley’s work on a set of American halls (Bradley, 1996). Complementary to these statements, Figs. 4b and 4c include the results of the percentage of data in each positional interval for this parameter, in steps of 1 JND in each interval, whereby only those results of the parameter for which there are significant values above 10% in all areas are included. These results correspond to the range from 0.10 to 0.30 for the stage without the orchestral shell and are similar to the conclusion drawn from Fig. 3b, and from 0.10 to 0.25 for the shell configuration. It is also worth noting that the black bar does not appear if the shell is present. 3.1.2. Late lateral level LJm A similar analysis was performed with the other monaural parameter associated with the subjective perception, known as the envelopment of sound for the subject, LJm. Thus Fig. 5a shows the experimental results obtained at the two source positions in all halls without the shell, and the averaged values of the parameter in the different positional areas. Mean values are very similar in all positional zones, and lower spatial dispersion is present at the centre of the stalls and in the vicinity of the side walls. Furthermore, it is worth noting that the range of variation of the parameter studied in all the rooms can be set from −6dB to 0 dB which is a narrower range of variation from the typical range specified in ISO 3382-1 for concert halls and multipurpose rooms with a volume <25,000 m3. For the completion of this study, and in order to normalize the distribution of data, Fig. 5b presents the percentage of the amount of data, in steps of 2 dB, relative to the total amount of data in each positional interval against these spatial intervals. It is worth noting that for the highest interval (−2dB, 0 dB) there is an increment of reception points from the centre of the halls to the lateral walls (black bar), except for the x/w > 1positions. Since the JND of this parameter remains unknown, the value of 2 dB S. Girón, T. Zamarre˜no, P. Bustamante – Objective Measures of Spatial Effects in Spanish Concert Halls 537 a) b) Fig. 5. a) Averaged late lateral level in the different intervals of normalized distance to the central axis of the halls: for all halls, the two source positions and without orchestra shell. The means and their standard deviation values are also shown, together with the typical range of values in this work and the typical range in ISO 3382-1; b) fraction of data for values of LJm in steps of 2 dB in the different intervals of x/w relative to the total amount of data in the same interval, for all halls without the orchestra shell on the stage. has been chosen due to the wide range of the variation: from −8.57 dB to +8.15 dB. In this representation, the results of the values of this parameter that fail to present a significant amount of data (above 10%) in all positional zones are ignored. By restricting the study to the 4 rooms in which acoustic measurements were taken in the two configurations, with and without the orchestra shell on the stage, set of Figs. 6, the averaged values remain constant in all positional areas and confirm the behaviour of the parameter values in all rooms studied in Fig. 5a, although slightly higher values are encountered. Both Figs. 5a and 6a, maintain the upper value of the range of variation of the parameter in these spaces (0 dB) despite the significant number of points that are ignored above 0 dB, because they are less than 10% in each 2 dB jump. Furthermore, these results relate mostly to receptors of MFA (this space has no orchestral shell) and are therefore not shown in Fig. 6a. In fact, the distribution of mean values is approximately −2dB when the shell is present and approximately −4dB in its absence (Fig. 6a). The gap between the two configurations tends to decrease for the areas closest to the side walls (x/w > 0.75). Regarding the spatial dispersion as assessed by the standard deviation, this is generally in the same order of magnitude for the two configurations of the stage, in all positional zones of the audience area. a) b) c) Fig. 6. a) Late lateral level results in the different intervals of normalized distance to the central axis of the halls for 4 halls, at S1 source position and for the two stage configurations. The mean values and their standard deviations for each interval are also shown, together with the typical range of values in this work and the ISO 3382-1 range; b) fraction of data for values of LJm in steps of 2 dB in the different intervals of x/w, relative to the total amount of data in the same interval, for 4 halls without the orchestra shell; c) idem with the orchestra shell on the stage. As for the complementary study conducted through Figs. 6b and 6c, the dataset with values of n/n∆xabove the 10% limit suggests shortening the LJm typical values to between −8dB and 0 dB for the without-shell configuration, in Fig. 5b the range is narrower (−6dB, 0 dB), and between −6dB and +2 dB for the configuration with the shell. The shell therefore increases the 544 Archives of Acoustics – Volume 37, Number 4, 2012 In all cases, the linear fits studied show very weak correlations. Taking into account the aforementioned assumptions, the only reliable correlations are those of (1-IACCEm)with A(with negative slope), (1IACCLm)with T, whereby the parameter increases when reverberation time increases, and that of LJm with V/N which also has a positive slope Eqs. (10)– (12). The values and linear regressions corresponding to the binaural parameters are plotted in Figs. 11a and b, respectively. LJm =−0.763(V/N) + 5.159, R2= 0.4861, P = 0.0081,(10) (1 −IACCEm) = −0.0001A+ 0.668, R2= 0.5538, P = 0.0035,(11) (1 −IACCLm) = 0.044T+ 0.759, R2= 0.5828, P = 0.0024.(12) In the same way as the simple image model indicates in concert halls and opera houses, the mean hall early lateral energy fractions are influenced by the hall width. Other pieces of work have found this correlation, for instance Gade (1989) concludes that JLFm has no meaningful connection to diffuse field theory a) b) Fig. 11. a) Hall-average values of (1-IACCEm)versus A; b) (1-IACCLm)values versus T, in the two configurations of the stages and their linear fits. but shows a highly significant relationship with hall width. Based on the data from all 32 halls, the correlation is R2= 0.43 and negative. However, if the analysis is restricted to 16 rectangular halls, it increases to R2= 0.67. In addition, the ratio 2H/W which determines the relative arrival time of early lateral and vertical reflections, as West (1966) suggested for concert halls, seems to be a geometrical measure in connection with early lateral fraction measures. In the halls studied in this work, the two geometrical magnitudes W, and 2H/W present a very poor relationship with the monaural fraction JLFm and with the binaural measure (1-IACCEm). Perhaps the low correlation in this work is due to the fact that orchestra shells provoke audible variations in the acoustic parameters, while the associated geometrical quantities remain unchanged. To sum up, under the large volume of data analyzed in this work and the many variables included it would be desirable to follow Bradley’s very recent suggestions (Bradley, 2011) in connection with carrying out experimental work of a more focused nature. 3.4. Summary and discussion In this section, a synopsis of the study is undertaken and the results are presented of the objective acoustic parameters, (spectrally averaged in accordance with the ISO 3382-1 standard (ISO, 2009)) that qualify the spatial impression in 9 performance halls in southern Spain (Andalusia) in the various approaches. The dependence of monaural and binaural parameters on source-receiver distance shows no predictable behaviour and experimental results have been compared with the theoretical calculations of Barron’s revised theory, which needs only the mean reverberation time and total hall volume for each hall (Subsec. 3.1, (Barron, Lee, 1988)). The study of spatial distribution of acoustic parameters in the rooms was jointly carried out by scaling the results at intervals of 1 JND (0.05 for JLFm and JLLFm, 0.075 for (1-IACCm)measures, and 2 dB for LJm, whose JND is not known) depending on the relative positions of the points of reception to the central axis of each room (for left and right-hand sides), which were normalized relative to half their respective mean width, (Subsec. 3.1). The orchestra shell commonly produces a small but perceptible effect on these parameters measured in the audience area, since for JLFm, Fig. 4a, the change may reach a value of 1 JND, especially in the areas closest to the side walls. In addition, the range of variation of JLFm in the set of values for all rooms is (0.10, 0.30) in the configuration without shell (Figs. 3a and 4a), and (0.10, 0.25) with the orchestral shell, Fig. 4a, which suggests that the shell favours a redistribution of the acoustic energy in the audience area in these performance spaces. Furthermore, the average values of JLFm in all positional S. Girón, T. Zamarre˜no, P. Bustamante – Objective Measures of Spatial Effects in Spanish Concert Halls 545 intervals are lower when the shell is present on stage than when it is absent. For the late lateral level LJm, however, the shell seems to favour an increase of the parameter, and in both configurations of the stage the range of variation is also much narrower than the typical range stated in ISO 3382-1 (ISO, 2009), (−8, 0) dB for the no-shell configuration, and (−6, 2) dB for the with shell, see set of Figs. 6. As for the binaural parameters, (Table 2), both the early lateral energy as well as the late energy exhibit similar spatial dispersion and are lower than the monaural parameters at the various positional intervals. For (1-IACCEm), the typical range in these spaces is (0.40, 0.70) for the two configurations of the stage. Parameter (1-IACCLm)exhibits the same behaviour in its mean values in each positional zone with a slight peak in the area between the central axis and side walls of the room (stalls zone). The standard deviations show that the spatial dispersions are not wide, and are slightly higher, generally, in the configuration without shell. The typical range of this parameter in these areas is (0.75, 0.90) regardless of whether there is an orchestra shell on stage. The effect on lateral energy parameters measured in the audience zone due to the change of source position on stage has been also studied in terms of the absolute value of the differences in their values, scaled at intervals of 1 JND at each reception point of each room. No predictable behaviour of these differences in the parameters associated with the variation of sourcereceiver distance, nor of the different laterality of the source sound position has been detected. In a statistical survey of all reception points, classified according to how they see the source in all rooms together, most of the seats have variations that are in the range 1–2 JND for the early lateral energy parameters, whilst for the late lateral energy, most points lie within differences <1JND (Figs. 7 and Table 3). Considering the behaviour of these differences for each room, JLFm presents very homogeneous values in MCT and LVT halls, as shown in Fig. 8a, which both have horseshoe typology. Generally for the late lateral level, LJm, Fig. 8b, a greater change is found in the values of these differences from place to place within all the halls studied. The results of the comparison between values of the two types of measures of lateral energy (monaural and binaural for the same temporal interval) obtained in the 9 halls in all the individual positions show that the correlations are very weak and that the relationships are not statistically significant, as shown in the set of Figs. 9. However, relationships improve when the data on each hall is considered exclusively, especially in certain halls, and when all hall-average data is studied for a comparison between JLFm and (1-IACCEm). In the latter case, the inclusion in the study of the results of a set of 15 halls from Okano’s work (Okano et al., 1998) further improves the correlation found (Fig. 10a). For the late lateral energy parameters, no similar data is available in the literature. However, by taking into account the values of the 9 halls studied in this work for the two configurations of the stage, the relationship between (1-IACCLm)and LJm parameters becomes similar to that obtained previously for the early lateral energy parameters as shown in Fig. 10b. Finally, the study in Subsec. 3.3.2 of the dependencies of these hall-average parameters with eight acoustic and geometrical parameters, (namely, mean width W, mean height-to-width ratio H/W, mean height-todepth ratio H/D, mean depth-to-width ratio D/W , total-volume-to-number-of-seats ratio V/N, audiencesurface-to-number-of-seats ratio SA/N, reverberation time T, and absorption A), enables the conclusion that the statistically significant relationships are the following linear regressions between the variables: (1-IACCEm)with A, Fig. 11a, with negative slope; (1-IACCLm)with T, Fig. 11b, with positive slope; and that of LJm with V/N, also with negative slope. It can be observed that, in all cases, the independent variable is related to the reverberation characteristic of the room (it must be taken into account that the major absorbent surface in these spaces is the audience zone, which is closely related to the V/N ratio). In this analysis, no other geometrical or acoustic variables have shown their potential influence on the values of the parameters related to spatial impression. 4. Conclusions Of the various aspects studied in this work on the monaural directional and binaural parameters (spectrally averaged in accordance with the ISO 3382-1 standard) to evaluate the spatial effects of the sound fields in 9 performance spaces in Andalusia (southern Spain), it is first worth emphasising that the dependence of the parameters on source-receiver distance in each hall has revealed that lateral acoustic energy is not distributed according to the theoretical predictions of the diffuse revised theory by Barron. As a result, a common methodology has been used for all areas and parameters under study. In this methodology the position of the receiving points are expressed in relation to their distance from the symmetry axis of the room (leftand right-hand sides) normalized with respect to half the average width of the room, and the acoustic parameters are scaled at intervals of their respective JND. The statistical treatment is the same for the parameters, data from all rooms with source in S1 and S2 without shell, and then the 4 rooms in which they are measured with the two configurations on the stage. The maintenance of conclusions when treating a wider statistical ensemble with the second case (smaller) gives validity to the results and conclusions when shells 546 Archives of Acoustics – Volume 37, Number 4, 2012 are on the stages. Statistical results show that the spatial distribution of early lateral energy fractions provides higher values for positions closer to the side walls (mean values for x/w > 1exceed those of the central area over 1 JND), whereas those parameters associated with late lateral energy and binaural parameters present a more uniform statistical distribution in the positional zones. The applied methodology has allowed, for each parameter studied, the systematic establishment of their characteristic ranges for the distribution of their values for the two shell configurations on the stage. These ranges are generally smaller than those suggested in ISO 3382-1. Statistical results have shown that the orchestral shell promotes a redistribution of acoustic energy into the audience area and produces a noticeable decrease of the early lateral energy fraction: this decrease is greater for positions close to the side walls. However, for the late lateral level the presence of the shell on stage leads to an increase in its values and a similarity in magnitude in all positional intervals. This suggests, at least initially, that shells focus sound by decreasing the initial lateral reflections, especially in the areas under balconies. This is reflected in the values of the limits of the ranges mentioned in the previous paragraph. The change in the two positions of the sound source on stage also influences these parameters; its influence is greater on the early lateral energy parameters, and the differences found due to the change are more homogeneous for the horseshoe typology. For the late lateral energy parameters, there is a lower incidence of change in position of the sound source and results present more uniformity in auditorium typologies. The analysis has not shown that the relative position of the receiver in front of the source (left/right) has a significant effect in any of the parameters analyzed. The change in position of the source implies that, for all the parameters analysed, more than 80% of receivers of the entire enclosures experience variations lower than 2 JND. For the (1-IACCLm)parameter, this percentage reaches 100% of receivers. Monaural and binaural parameters describe very different physical characteristics of the sound field, and thus correlations are very weak when studying pairs of results at each position of signal reception. Exclusively when each hall is described by a single value (the spatial average) an acceptable relationship between the parameters is found both for the early and late timeinterval parameters. Finally, in order to draw conclusions in relation to design variables, monaural and binaural parameters have been compared with the mean values of geometrical and acoustical variables of the performance spaces. Regressions show that neither the mean width, nor the mean-height/mean-width ratio is significant and that in all cases, parameters related to reverberation are the key (absorption, reverberation time, and V/N ratio). Acknowledgment The authors wish to express their gratitude to the staff and management of each hall for their cooperation in allowing the measurements to be carried out. This work has been sponsored by FEDER funds and the Spanish Ministry of Science and Technology (MCYT) within the projects with references BIA2003-09306CO4-02, and BIA2010-20523. References 1. Abdou A., Guy R.W. (1996), Spatial information of sound fields for room-acoustics evaluation and diagnosis, Journal of the Acoustical Society of America, 100, 3215–3226. 2. Ando Y. (2007), Handbook of Acoustics, Thomas Rossing, Editor, Springer-Verlag, New York Chapter 10. 3. Barron M. (1971), The subjective effects of first reflections in concert halls - the need for lateral reflections, Journal of Sound and Vibration, 15, 475–494. 4. Barron M. (2001), Late lateral energy fractions and the envelopment question in concert halls, Applied Acoustics, 62, 185–202. 5. Barron M., Lee L-J. (1988), Energy relations in concert auditoria I, Journal of the Acoustical Society of America, 84, 618–628. 6. Barron M., Marshall A.H. (1981), Spatial impression due to early lateral reflections in concert halls. The derivation of a physical measure, Journal of Sound and Vibration, 77, 211–232. 7. Beranek L.L. (1962), Music, Acoustics and Architecture, John Wiley & Sons, New York. 8. Beranek L.L. (1992), Concert hall acoustics, The Journal of the Acoustical Society of America, 92, 1– 39. 9. Blauert J., Lindemann W. (1986a), Auditory spaciousness: Some further psychoacoustic analyses, The Journal of the Acoustical Society of America, 80, 533– 542. 10. Blauert J., Lindemann W. (1986b), Supplementary psychoacoustical results on auditory spaciousness, Acustica, 59, 292–293. 11. Bradley J.S. (1994), Comparison of concert hall measurements of spatial impression, Journal of the Acoustical Society of America, 96, 3525–3535. 12. Bradley J.S. (1996), Some effects of orchestra shells, Journal of the Acoustical Society of America, 100, 889898. 13. Bradley J.S. (2011), Review of objective room acoustics measures and future needs, Applied Acoustics, 72, 713–720. S. Girón, T. Zamarre˜no, P. Bustamante – Objective Measures of Spatial Effects in Spanish Concert Halls 547 14. Bradley J.S., Reich R.D., Norcross S.G. (2000), On the combined effects of earlyand late-arriving sound on spatial impression in concert halls, Journal of the Acoustical Society of America, 108, 651–661. 15. Bradley J.S., Soulodre G.A. (1995a), The influence of late-arriving energy on spatial impression, Journal of the Acoustical Society of America, 97, 2263– 2291. 16. Bradley J.S., Soulodre G.A. (1995b), Objective measures of listener envelopment, Journal of the Acoustical Society of America, 98, 2590–2597. 17. Cox T.J., Davies J., Lam Y.W. (1993), The sensitivity of listeners to early sound field changes in auditoria, Acustica, 79, 27–41. 18. Cremer L. (1989), Early reflections in some modern concert halls, Journal of the Acoustical Society of America, 85, 1213–1225. 19. Damaske P., Ando Y. (1972), Inter aural cross correlation for multichannel loudspeaker reproduction, Acustica, 27, 232–238. 20. De Vries D., Hulsebos E.M., Baan J. (2001), Spatial fluctuations in measures for spaciousness, Journal of the Acoustical Society of America, 110, 947–954. 21. Evjen P., Bradley J.S., Norcross S.G. (2001), The effect of late reflections from above and behind on listener envelopment, Applied Acoustics, 62, 137–153. 22. Gade A.C. (1989), An acoustical survey of eleven European concert halls - a basis for discussion of halls in Denmark, The Acoustics Laboratory, Tech. Univ. of Denmark, [Report No.44]. 23. Girón S., Galindo M., Zamarre˜ no T. (2008), Distribution of lateral acoustic energy in Mudejar-Gothic churches, Journal of Sound and Vibration, 315, 1125– 1142. 24. Girón S., Zamarre˜ no T., Galindo M. (2010), Experimental study of support parameters in auditorium and proscenium stages, Acta Acustica united with Acustica, 96, 1026–1041. 25. Hidaka T., Beranek L.L., Okano T. (1995), Interaural cross-correlation, lateral fraction, and lowand high-frequency sound levels as measures of acoustical quality in concert halls, Journal of the Acoustical Society of America, 98, 988–1007. 26. International Organization For Standardization ISO 3382-1:2009(E), (2009), Acoustics-Measurement of room acoustic parameters, part 1: Performance rooms, Geneva, Switzerland. 27. Kleiner M. (1989), A new way of measuring the lateral energy fractions, Applied Acoustics, 27, 321–327. 28. León A.L., Sendra J.J., Navarro J., Zamarre- ˜ no T. (2007), Acoustics and restoration in theatres of Andalucia [in Spanish: Ac´ustica y Rehabilitación en Teatros de Andaluc´ıa], Secretariado de Publicaciones de la Universidad de Sevilla, Sevilla. 29. Marshall A.H. (1967), A note on the importance of room cross-section in concert halls, Journal of Sound and Vibration, 5, 100–112. 30. Marshall A.H. (1968), Levels of reflection masking in concert halls, Journal of Sound and Vibration, 7, 116–118. 31. MIREM (Mapa Informatizado de Recintos Esc´enicos y Musicales; Fundación Autor, Sociedad General de Autores y Editores) http://www.artenetsgae.com- /mire/index.htm [Digital Scenic and Musical Enclosure Map, Author Foundation] (Accessed September 2012, currently only in Spanish). 32. Morimoto M., Iida K., Sakagami K. (2001), The role of reflections from behind the listener in spatial impression, Applied Acoustics, 62, 109–124. 33. Morimoto M., Nakagawa, K. Iida K. (2008), The relation between spatial impression and the law of the first wavefront, Applied Acoustics, 69, 132-140. 34. Okano T., Beranek L.L., Hidaka T. (1998), Relations among interaural cross-correlation coefficient (IACCE), lateral fraction (LFE), and apparent source width (ASW) in concert halls, Journal of the Acoustical Society of America, 104, 255–265. 35. Peutz V.M.A. (1971), Articulation loss of consonants as a criterion for speech transmission in a room, Journal of the Audio Engineering Society, 19, 915–919. 36. Sabine W.C. (1993), Collected Papers on Acoustics. (Cambridge: Harvard University Press), Reprinted by Acoustical Society of America, New York. 37. Schroeder M.R., Atal B.S., Sessler G.M., West J.E. (1966), Acoustical measurements in the Philharmonic Hall (New York), Journal of the Acoustical Society of America, 40, 434–440. 38. Soeta Y., Nakagawa S., Tonoike M., Ando Y. (2002), Magnetoencephalographic responses corresponding to individual subjective preference of sound fields, Journal of Sound and Vibration, 258, 419–428. 39. West J.E. (1966), Possible subjective significance of the ratio of height to width of concert halls, Journal of the Acoustical Society of America, 40, 1245.