The Acoustic Environment of York Minster’s Chapter House
The Acoustic Environment of York Minster’s Chapter House
Álvarez-Morales, Lidia;Lopez, Mariana;Álvarez-Corbacho, Ángel
2020-01-30 00:00:00
acoustics Article The Acoustic Environment of York Minster’s Chapter House 1 , 1 2 Lidia Álvarez-Morales * , Mariana Lopez and Ángel Álvarez-Corbacho Department of Theatre, Film, Television and Interactive Media, University of York, York YO10 5GB, UK; mariana.lopez@york.ac.uk Instituto Universitario de Arquitectura y Ciencias de la Construcción, Escuela Técnica Superior de Arquitectura, Universidad de Sevilla, 41012 Seville, Spain; arqangel@us.es * Correspondence: lidia.alvarezmorales@york.ac.uk Received: 13 December 2019; Accepted: 27 January 2020; Published: 30 January 2020 Abstract: York Minster is the largest medieval Gothic cathedral in Northern Europe, renowned for its magnificent architecture and its stained glass windows. Both acoustic measurements and simulation techniques have been used to analyse the acoustic environment of its Chapter House, which dates from the 13th-century and features an octagonal geometry with Gothic Decorated stone walls replete of geometric patterns and enormous stained glass windows, covered by a decorated wooden vault. Measured and simulated room impulse responses served to better understand how their architectural features work together to create its highly reverberant acoustic field. The authors start by analysing its acoustic characteristics in relation to its original purpose as a meeting place of the cathedral’s Chapter, and end by reflecting on its modern use for a variety of cultural events, such as concerts and exhibitions. This work is part of the “Cathedral Acoustics” project, funded by the EC through the Marie-Sklodowska-Curie scheme. Keywords: heritage acoustics; cathedral acoustics; room acoustics; York Minster; acoustic simulation 1. Introduction Research on the acoustics of heritage sites allows us, firstly, to understand and preserve the acoustic behaviour of spaces threatened by the passage of time, or modified by human intervention, by capturing the acoustic information digitally in the form of room impulse responses (RIR). Secondly, research in the field furthers our understanding on how acoustic conditions have changed over time, opening up discussions on what such changes might have meant to our ancestors and their experience of such sites. This paper focuses on assessing the acoustics of the Chapter House of the Cathedral and Metropolitical Church of St Peter in York, better known as the York Minster. York Minster is the largest medieval Gothic cathedral in Northern Europe, renowned for its outstanding architecture. Its Chapter House is an independent venue attached to the north transept of the cathedral, in which a great variety of events are held on a regular basis. The York Minster ’s Chapter House is considered a remarkable piece of architecture itself and its acoustics a major feature of its character. This work provides new data about the acoustics of the York Minster ’s Chapter House, that can be used to preserve its acoustic field as well as be used as a starting point for further research on more specific aspects of the sonic environment. This research contributes to the existing literature on acoustics of cathedrals and heritage buildings, using this space as a representative example of polygonal-shaped English Chapter Houses to better understand how their proportions, verticality, rich decoration and hard finishing materials work together to create their reverberant acoustic field. Acoustics 2020, 2, 13–36; doi:10.3390/acoustics2010003 www.mdpi.com/journal/acoustics Acoustics 2020, 2 14 The experimental study of the acoustics of the Chapter House was conducted following the well-established methodology outlined in ISO 3382-part 1 [1], as well as additional guidelines developed for churches and cathedrals [2,3]. As in previous studies on cathedral acoustics [4–6], a thorough analysis of the acoustic behaviour of the space was conducted through the acoustic parameter values derived from the measured RIR, registered at a set of positions which reflected the wide diversity of uses of the site. The sound reflection pattern in polygonal-shaped rooms is always a challenge because of the unwanted echoes, low frequency colouration, and non-diusion conditions commonly found in those spaces [7]. The octagonal shape of the Chapter House, which has multiple parallel walls, is one of those shapes. Therefore, a careful inspection of the reflection pattern and the early arriving energy at each reception point has been conducted. The reflection pattern and the sound decay curves derived from the measured RIR has been analysed in previous studies of cathedrals, as their rich architectural features and complex forms sometimes give rise to particular acoustic phenomena such as coupled spaces or singular patterns of early reflections [8,9]. In addition to capturing RIR on site, the study of the Chapter House utilised computer modelling to further our understanding of the space and its use. Three-dimensional reconstruction and acoustic simulation techniques present a powerful tool to recreate the acoustic field within a space, helping us to deepen the analysis of its acoustic behaviour, while also oering the opportunity to acoustically experience sites that no longer exist through the process of auralisation [10,11]. Simulation software tools based on geometrical acoustics (GA) algorithms have been also successfully used in similar buildings before for a great variety of purposes, for instance to explore dierent intervention options and adapt their acoustics to a specific use [12] and to assess how their acoustics change in an occupied state [13]. In this particular case, working with the model of the Chapter House allowed the research team to analyse in detail the influence of the polygonal-shape of the room on the early reflection pattern. Such analysis was done by considering the dierent configurations of the space depending on its uses. Furthermore, simulations were used to give an insight on the acoustics of the Chapter House in its original state, before any restoration work or intervention was made. Carrying the on-site acoustic measurements prior to the simulation work is crucial to the validation of the acoustic model created, first, to represent the current state of the building, and then, modified as required. However, it is important to acknowledge that the main challenges in research on acoustic simulations of heritage sites are the uncertainties on the acoustic properties of historical materials, the modelling of certain sound sources and the inherent limitations of the geometrical acoustic theory itself in which the simulation software tools are based on [14]. 2. The York Minster’s Chapter House 2.1. Overview of its Construction Process and Architectural Description York Minster ’s history dates back to the year 627, when a wooden church dedicated to St Peter was built [15]. The construction of the medieval cathedral first began around 1225, with the construction of the transept on the foundations of the Norman cathedral, which was demolished in stages as Gothic additions were made [16]. Its constructive evolution continued throughout more than 250 years, until the north and south towers were built. The Minster was consecrated in 1472, a few years before its completion [17]. The cathedral of York is considered one of the finest medieval buildings in Europe, representing almost every stage of the Gothic style of architecture from 1230 to 1475 and having the highest proportion of medieval stained glass of any European cathedral [18]. Not surprisingly, the York Minster is one of the most-visited buildings in Britain [19]. The Chapter House has a polygonal shape, which is a peculiarity of English thirteenth-century architecture, not found in any other place in Europe (with few exceptions such as the Chapter House in Elgin cathedral (octagonal shape, 1270) in Scotland). There is evidence for 25 polygonal Chapter Houses in England, but more than half of them no longer exist [20]. There is no certainty as to the reasons Acoustics 2020, 2 15 behind the popularity of the polygonal form, although previous research has linked the choice of shape to acoustics [21,22], as a way to avoid the “intolerable resonance” of the rectangular chapter houses [23]. Nevertheless, Wickham [23] also stated, “the resonance of some of the polygonal houses, both of those with central pillars and of those without them, is also intolerable”, and suggested that the polygonal shape emerged to serve its original purpose as a meeting place where all canons were considered equals, intended Acoustics 2020, 2 15 to avoid the dean sitting at the head as in the traditional rectangular Chapter Houses. Nevertheless, it seems more plausible that their shape, decoration and proportions derived from the tendency in English chapter houses [23]. Nevertheless, Wickham [23] also stated, “the resonance of some of the polygonal architecture around the year 1300 of “compiling designs for full-scale buildings or parts of buildings out houses, both of those with central pillars and of those without them, is also intolerable”, and of enlargements of the microarchitecture from the portals of thirteenth-century French cathedrals” [24], suggested that the polygonal shape emerged to serve its original purpose as a meeting place where where the wall-framework structure, the geometric uniformity and the use of large windows had a all canons were considered equals, intended to avoid the dean sitting at the head as in the traditional profoundly visual value, being a prestigious architectural statement by the cathedral’s canons. rectangular Chapter Houses. Nevertheless, it seems more plausible that their shape, decoration and Regardless of the reasons behind the octagonal shape of the York Chapter House, its architecture proportions derived from the tendency in English architecture around the year 1300 of “compiling is remarkable, designs for having full-scal no e bu central ildingpillar s or pa supporting rts of buildithe ngs ou vaulting t of enceiling, largemewhich nts of th iseinstead microarsupported chitecture by a spir fre-like om the port timber als of thi roof str rteenth- ucture. century French ca It is considered, thedr along als” [24], wh with the ere the octagon wall-fr in a Ely mework Cathedral, structure, one of the geometric uniformity and the use of large windows had a profoundly visual value, being a the masterpieces of English medieval carpentry [25]. Although there is a lack of documentary proof prestigious architectural statement by the cathedral’s canons. related to the construction of the Chapter House [26], some evidence suggest that its construction Regardless of the reasons behind the octagonal shape of the York Chapter House, its architecture process was completed around the last two decades of the 13th century [15]. Work started right after is remarkable, having no central pillar supporting the vaulting ceiling, which is instead supported by the current transepts were erected, and it was completed before the current nave was built. The space a spire-like timber roof structure. It is considered, along with the octagon in Ely Cathedral, one of the is built in Early Decorated Gothic style, so-called geometric style (1250–90) in English architecture, masterpieces of English medieval carpentry [25]. Although there is a lack of documentary proof being designed to promote verticality and with the clear purpose of creating visual impact. The York related to the construction of the Chapter House [26], some evidence suggest that its construction Minster ’s Chapter House took the Westminster Chapter House as a source of inspiration, being the process was completed around the last two decades of the 13th century [15]. Work started right after one in Westminster designed not only for the Chapter meetings but also to be used by Henry III [27], the current transepts were erected, and it was completed before the current nave was built. The space fact that also may explain its exuberance. is built in Early Decorated Gothic style, so-called geometric style (1250–90) in English architecture, being de As can be signe seen d toin promot the floor e vert plan icalitincluded y and within thFigur e clear purpose o e 1, the ground f creating layout visual of impact. The York the Chapter House Minster’s Chapter House took the Westminster Chapter House as a source of inspiration, being the is a quasi-regular octagon. It is annexed to the cathedral in its north transept through an L-shaped one in Westminster designed not only for the Chapter meetings but also to be used by Henry III [27], vestibule, also built in the Early Decorated style. The building is perfect in proportions, being about as fact that also may explain its exuberance. high as its axis is long. For more details about its dimensions see Table 1. Figure Figure 1. 1. Floor Floor plan plan of the York Minster’s Ch of the York Minster ’s Chapter apter House House wi with th the sou the sour rce (S ce ) and recei (S) and receiver ver (R) positions (R) positions set for the acoustic measurements. The complete floor plan of the Minster is included for showing its set for the acoustic measurements. The complete floor plan of the Minster is included for showing location. its location. Table 1. Geometrical data of the York Minster ’s Chapter House. As can be seen in the floor plan included in Figure 1, the ground layout of the Chapter House is a quasi-regular octagon. It is annexed to the cathedral in its north transept through an L-shaped Approximate Max Interior Max Interior Octagon Side’s Approximate Interior vestibule, also built in the Early Decorated style. The building is perfect in proportions, being about Interior Volume Height Length Length Floor Surface as high as its axis is long. For more details about its dimensions see Table 1. 3 a 2 5280 m 20.3 m 19.2 m 6.95 m 240 m Measured glass to glass. Table 1. Geometrical data of the York Minster’s Chapter House. Approximate interior Max interior Max interior Octagon side’s Approximate interior Volume Height Length Length floor Surface 3 2 5,280 m 20.3 m 6.95 m 240 m 19.2 m Measured glass to glass. Acoustics 2020, 2 16 Acoustics 2020, 2 16 The The Chapter House Chapter House is is accessed accessed by the w by the west-side est-side of of the octa the octagon gon ((see see Fi Figur gure e11 ), through a ), through a rounded rounded archway divided by a pillar with a statue of The Virgin and Child [28]. The oak doors at each side of archway divided by a pillar with a statue of The Virgin and Child [28]. The oak doors at each side of the entrance date from the 13th century, and their original ironwork rich decoration is still visible the entrance date from the 13th century, and their original ironwork rich decoration is still visible [29]. [29]. The current polychrome Minton tiled floor (see Figure 2) is a result of the restoration of 1843-1845. The current polychrome Minton tiled floor (see Figure 2) is a result of the restoration of 1843- There is no feasible trace of medieval floor in the Chapter House, which is thought to have been 1845. There is no feasible trace of medieval floor in the Chapter House, which is thought to have been made of stone [30], although it is also possible that the decorative pattern tiles that still survive at the made of stone [30], although it is also possible that the decorative pattern tiles that still survive at the Westminster Chapter House once existed at York [16]. Eighteenth century views show an evenly paved Westminster Chapter House once existed at York [16]. Eighteenth century views show an evenly floor, consisting of rectangular tiles with no visible decorative pattern [31]. paved floor, consisting of rectangular tiles with no visible decorative pattern [31]. Figure 2. York Minster ’s Chapter House interior views: looking to the West side (left) and looking to Figure 2. York Minster’s Chapter House interior views: looking to the West side (left) and looking to the East side (right). the East side (right). Seven of the Chapter House’s sides have a uniform design (the eighth side is the entrance). Seven of the Chapter House’s sides have a uniform design (the eighth side is the entrance). The The lower part of the room is organised as a peripteral colonnade, in which a constant pattern of black lower part of the room is organised as a peripteral colonnade, in which a constant pattern of black Purbeck marble columns separates the canon stalls, 6 in each side (apparently, the six stalls on the East Purbeck marble columns separates the canon stalls, 6 in each side (apparently, the six stalls on the side were designated for the Dean and dignitaries [30]). The stalls are covered by limestone canopies East side were designated for the Dean and dignitaries [30]). The stalls are covered by limestone which are highly decorated with a variety of forms and carves, from humorous faces to animals, which canopies which are highly decorated with a variety of forms and carves, from humorous faces to were originally coloured [30]. The upper part of the canopies forms a passage surrounding the room animals, which were originally coloured [30]. The upper part of the canopies forms a passage in front of the windows, with the exception of the doorway. Magnificent stained glasses (framed by surrounding the room in front of the windows, with the exception of the doorway. Magnificent limestone clusters of shafts or nerves) dominate the upper space. On the access side, the limestone stained glasses (framed by limestone clusters of shafts or nerves) dominate the upper space. On the tabernacle over the door contained coloured figures (possibly of Christ and the 12 Apostles) lost by the access side, the limestone tabernacle over the door contained coloured figures (possibly of Christ and 17th century [32], and the blind limestone window above had a series of polychromy medieval wall the 12 Apostles) lost by the 17th century [32], and the blind limestone window above had a series of paintings, removed during restoration in 1840 [33]. polychromy medieval wall paintings, removed during restoration in 1840 [33]. In the exterior, the structure of the Chapter House is reinforced by buttresses on the octagon In the exterior, the structure of the Chapter House is reinforced by buttresses on the octagon corners. Inwardly, these buttresses become pillars on which the structure of the vault rests. The current corners. Inwardly, these buttresses become pillars on which the structure of the vault rests. The ribbed vault, divided into sixteen segments, is a fine plaster and lath work dated from 1797, when the current ribbed vault, divided into sixteen segments, is a fine plaster and lath work dated from 1797, 14th century wood ceiling panels were replaced due to deterioration [31]. The original panels (some when the 14th century wood ceiling panels were replaced due to deterioration [31]. The original now stored in the Minster ’s crypt) were decorated with paintings of kings and bishops and the bosses panels (some now stored in the Minster’s crypt) were decorated with paintings of kings and bishops wer and the e cover bosse ed with s were silver covere [34], d with fillingsi the lver spaces [34], fi outside lling the sp the hexadecagon. aces outside the hexa The entir deca e ceiling gon. The enti was again re restor ceiling w ed in as 1976 again when restored the curr in 19 ent 76 when decoration the current was applied. decoration was applied. Acoustics 2020, 2 17 2.2. A Singular Space for A Variety of Uses As any other Chapter House, this space was constructed to provide a large and decorous place to host the regular meetings of the Chapter for the daily business of the cathedral. Forty-four seats were placed around the space as described in the previous section, a number greater than the canons in York at the time of the construction [16]. For most Chapter meetings at that time, only a minority of the stalls were occupied [26], for example, in 1310 only three canons met in the Chapter House to arrange a date for the election of a new dean [35]. However, it has a long history of being used for events other than Chapter ’s meetings, for example, in 1297, when the government of England was York-based, the Parliament of Edward I, king of England, met in the Chapter House in October [36]. The space is still used for the meetings of the Corporate body of the York Minster. The College of Canons (32 canons), use the space regularly. The current Chapter is made up by 9 members, but they th do not always meet there. According to Harrison [29], “at the beginning of the 20 century some of the meetings were moved to the Zouche chapel, due to the diculties to heat the space and because of the echo which every sound made in it produces”. Nowadays, the configuration of the space changes depending on the meeting [37]. Such changes are not only in terms of the location of participants, but also, depending on the occasion, temporary pieces of furniture are used, mainly wooden tables and wooden or leather Victorian-style chairs [20] and the floor is sometimes covered with removable carpets. J. Archuleta, verger at the York Minster, shared some insight on the matter [38], “sometimes the stone seats around the space are used, perhaps one speaker at a time may stand towards the centre, and other times, there is a table placed in the centre of the Chapter House and modern chairs are used”. Contemporary uses of the Chapter House go beyond meetings, and include concerts, talks, tours and other social events held on a regular basis. The venue can also be hired to host dinners and drink receptions. L. Power, head of events and learning at the York Minster, set the capacity of the room at about 200 people, and when concerts are taking place it is a regular occurrence for the space to be fully booked [39], which mean approximately 130 spectators. Temporary stages, choir stalls, and audience seats (normally light weight upholstery portable chairs) are used as needed, and ephemeral solutions are sometimes included to meet the acoustic requirements of each use. For example, there is no permanent electroacoustic system installed in the Chapter House, but a portable system is used when needed, according with both Archuleta and Power [38,39]. Figure 3 shows the usual arrangement patterns between performers/sound sources and listeners used in the Chapter House. Figure 3a depicts the “front stage” arrangement; Figure 3b represents what is called “the amphitheatre-like arrangement” [40], and Figure 3c shows the original arrangement, in which the stone stalls are used both for performers and listeners. It is important to bear in mind that these arrangements are very flexible, and the configuration of the space is adapted to the requirements of each event. For instance, the audience area in “a” and “b” could expand to the stalls; in “c” the performers could be moved to any of the others sets of stalls depending on the location occupied by the person talking in each moment during the meeting. Configurations “a” and, occasionally, “b” are typically used for concerts and music events, and configurations “a”, “b” and “c” are used for meetings and talks. Figure 3 does not include the configurations used in events like tours or dinners. Acoustics 2020, 2 18 Acoustics 2020, 2 18 Figure 3. Examples of some typical configurations of the Chapter House: (a) front stage arrangement; (b) amphitheatre-like arrangement; and (c) original surrounding arrangement. Figure 3. Examples of some typical configurations of the Chapter House: (a) front stage arrangement; (b) amphitheatre-like arrangement; and c) original surrounding arrangement. 3. Methodology of Analysis 3. Methodology of Analysis The acoustical study of the Chapter House of the York Minster was based on acoustic measurements and simulation techniques. The room impulse responses (RIR) registered on site serve to assess the The acoustical study of the Chapter House of the York Minster was based on acoustic objectives acoustic characteristics of the space. Measured RIR are also used as a reference to adjust the measurements and simulation techniques. The room impulse responses (RIR) registered on site serve acoustic model in order to ensure that the computational model is reliable in its representation of the to assess the objectives acoustic characteristics of the space. Measured RIR are also used as a reference soundfield, enabling conclusions to be drawn from the simulations as well as making it possible for to adjust the acoustic model in order to ensure that the computational model is reliable in its simulated RIR to be used for auralisations that plausibly represent the acoustic environment of the representation of the soundfield, enabling conclusions to be drawn from the simulations as well as space under dierent conditions. Simulation techniques are also utilised to assess the acoustic impact making it possible for simulated RIR to be used for auralisations that plausibly represent the acoustic of restoration works, as well as dierent configurations related to the variety of uses of the Chapter environment of the space under different conditions. Simulation techniques are also utilised to assess House throughout history. the acoustic impact of restoration works, as well as different configurations related to the variety of uses of the Chapter House throughout history. 3.1. Acoustic Measurements in the Chapter House The complete RIR measurement session involved the entire cathedral, including seven sound 3.1. Acoustic Measurements in the Chapter House source positions selected by considering the architectural characteristics and the variety of uses of The complete RIR measurement session involved the entire cathedral, including seven sound the space. Moreover, 65 receiver positions were located throughout the dierent audience areas. source positions selected by considering the architectural characteristics and the variety of uses of the The number of source-receiver combinations characterised was selected in order to have representative space. Moreover, 65 receiver positions were located throughout the different audience areas. The data of each of the “subspaces” in the cathedral. This article focuses on the measurements and number of source-receiver combinations characterised was selected in order to have representative simulations linked to the Chapter House, and no acoustic information is given about the other data of each of the “subspaces” in the cathedral. This article focuses on the measurements and subspaces, which will be analysed in future articles. simulations linked to the Chapter House, and no acoustic information is given about the other The acoustic measurements were carried out under unoccupied conditions taking ISO3382-1 [1] subspaces, which will be analysed in future articles. as a reference, as well as considering the specific guidelines developed for similar buildings [2,3]. The acoustic measurements were carried out under unoccupied conditions taking ISO3382-1 [1] The signal utilised was a 24-second-long sine-sweep emitted through a dodecahedral loudspeaker as a reference, as well as considering the specific guidelines developed for similar buildings [2,3]. The with a frequency response from 50 Hz to 10 kHz (NTi DS3 loudspeaker together with a NTi PA3 power signal utilised was a 24-second-long sine-sweep emitted through a dodecahedral loudspeaker with a amplifier) at a height of 1.5 m. The level and the duration of the excitation signal were set to ensure a frequency response from 50 Hz to 10 kHz (NTi DS3 loudspeaker together with a NTi PA3 power reliable calculation of T which means that an impulse-to-noise ratio (INR) higher than 45 dB in all 30, amplifier) at a height of 1.5 m. The level and the duration of the excitation signal were set to ensure a octave bands would be achieved, even in those receivers located at the furthest positions from the reliable calculation of T30, which means that an impulse-to-noise ratio (INR) higher than 45 dB in all source (see Figure 4). octave bands would be achieved, even in those receivers located at the furthest positions from the source (see Figure 4). Acoustics 2020, 2 19 Acoustics 2020, 2 19 125 250 500 1000 2000 4000 Frequency (Hz) S-R individual combinations Averaged value Minimum required Figure 4. Impulse-to-noise ratio (INR) measured values obtained at each source-receiver combination Figure 4. Impulse-to-noise ratio (INR) measured values obtained at each source-receiver combination characterised in the York Minster ’s Chapter House (grey lines) and the spatially averaged values (green characterised in the York Minster’s Chapter House (grey lines) and the spatially averaged values line). The minimum value of INR required for a reliable calculation of T is also shown (red line). (green line). The minimum value of INR required for a reliable calculat 30 ion of T30 is also shown (red line). The loudspeaker was placed in two dierent source positions inside the Chapter House: one on the symmetry axes of the space, in the East side, which is considered a common position for The loudspeaker was placed in two different source positions inside the Chapter House: one on performers during cultural activities (Front stage configuration, Figure 3a) (S1); and another one at the the symmetry axes of the space, in the East side, which is considered a common position for centre of the octagon (Amphitheatre-like configuration, Figure 3b) (S2). Up to 11 receiver positions performers during cultural activities (Front stage configuration, Figure 3a) (S1); and another one at were distributed throughout the space from where a set of RIR were captured by using a B-format the centre of the octagon (Amphitheatre-like configuration, Figure 3b) (S2). Up to 11 receiver microphone (Soundfield ST450) and a binaural dummy head (Neumann KU 100). Table 2 summarises positions were distributed throughout the space from where a set of RIR were captured by using a the source and receiver combinations included in the survey. The exact positions of S1 and S2 and the B-format microphone (Soundfield ST450) and a binaural dummy head (Neumann KU 100). Table 2 receiver points can be seen in Figure 1. summarises the source and receiver combinations included in the survey. The exact positions of S1 The digital audio workstation Pro Tools 12 and MATLAB software were used to capture and and S2 and the receiver points can be seen in Figure 1. process the signals respectively. Then two commercial software tools, WinMLS2004 and IRIS 1.4, were The digital audio workstation Pro Tools 12 and MATLAB software were used to capture and employed for the acoustic analysis of the RIR and the calculation of the acoustic parameters. process the signals respectively. Then two commercial software tools, WinMLS2004 and IRIS 1.4, Environmental conditions were monitored during the session, registering an average temperature were employed for the acoustic analysis of the RIR and the calculation of the acoustic parameters. of 18 C and 60% of relative humidity. Environmental conditions were monitored during the session, registering an average temperature of 18 °C and 60% of relative humidity. Table 2. Summary of the source-receiver combinations included in the survey. Table 2. Summary of the source-receiver combinations included in the survey. Source position ID S1 S2 Num. Rec. points 10 8 Source position ID S1 S2 Receiver points ID R01 – R10 R01, R04 – R06, R08 – R11 Num. Rec. points 10 8 Receiver points ID R01 – R10 R01, R04 – R06, R08 – R11 3.2. Acoustic Simulation of the Space 3.2. Acoustic Simulation of the Space The objective evaluation and validation of the acoustic simulation was based on the comparison of the room acoustic parameters derived from the simulation and those calculated from the RIR registered The objective evaluation and validation of the acoustic simulation was based on the comparison in the space. The acoustic simulation was carried out with Odeon 14.0, by using the precision algorithm. of the room acoustic parameters derived from the simulation and those calculated from the RIR The IR length and the number of late rays were manually set to 8 ms and 150,000 rays respectively, registered in the space. The acoustic simulation was carried out with Odeon 14.0, by using the while the early reflection settings were automatically set (8976 of early rays with a transition order of precision algorithm. The IR length and the number of late rays were manually set to 8 ms and 150,000 2). The geometrical model was imported into the acoustic software from the SketchUp modelling tool, rays respectively, while the early reflection settings were automatically set (8,976 of early rays with a which was used to build the geometrical model. transition order of 2). The geometrical model was imported into the acoustic software from the SketchUp modelling tool, which was used to build the geometrical model. 3.2.1. Acoustic Model The 3D model was built from architectural plans and laser measurements taken on site, by using the modelling software tool SketchUp. The simplified model was created following the guidelines for INR (dB) Acoustics 2020, 2 20 Acoustics 3.2.1. Acou 2020,st 2 ic Model 20 The 3D model was built from architectural plans and laser measurements taken on site, by using the modelling software tool SketchUp. The simplified model was created following the guidelines geometrical acoustics (GA) algorithms, avoiding overly detailed surfaces and too complex forms [41]. for geometrical acoustics (GA) algorithms, avoiding overly detailed surfaces and too complex forms The first significant simplification is the modelling of the highly decorated limestone canopies as a [41]. The first significant simplification is the modelling of the highly decorated limestone canopies smooth surface and the simplification of the upper passage. To compensate for the limestone area lost as a smooth surface and the simplification of the upper passage. To compensate for the limestone with this simplification, the acoustic absorption coecient assigned to this layer in the acoustic model area lost with this simplification, the acoustic absorption coefficient assigned to this layer in the was slightly increased. This is a common practice when modelling complex buildings, in which rich acoustic model was slightly increased. This is a common practice when modelling complex buildings, decorative patterns can make hard surfaces acoustically more absorbent [42]. The scattering coecient in which rich decorative patterns can make hard surfaces acoustically more absorbent [42]. The was estimated taking into consideration the surface’s depth variations [43]. scattering coefficient was estimated taking into consideration the surface’s depth variations [43]. The second simplification, is the definition of the limestone nerves that frame the stained glasses. The second simplification, is the definition of the limestone nerves that frame the stained glasses. In this regard, three dierent possibilities were tested: no nerves, 2D nerves and 3D nerves. Omitting In this regard, three different possibilities were tested: no nerves, 2D nerves and 3D nerves. Omitting the nerves implied an increment of stained glasses surface of about 120 m at the expense of the stone the nerves implied an increment of stained glasses surface of about 120 m at the expense of the stone surface, so, considering the dierences in the acoustic properties between both materials, this option surface, so, considering the differences in the acoustic properties between both materials, this option was discarded. Acoustic models including the 2D nerves and 3D nerves were tested in the simulation was discarded. Acoustic models including the 2D nerves and 3D nerves were tested in the simulation software. Modelling the nerves in 3D resulted in an increment of 200 m of limestone distributed in software. Modelling the nerves in 3D resulted in an increment of 200 m of limestone distributed in 1887 planes added to the model including the 2D nerves. After analysing the percentage of surface 1,887 planes added to the model including the 2D nerves. After analysing the percentage of surface included in each configuration and considering the costs of the calculation process of each model, the included in each configuration and considering the costs of the calculation process of each model, the option of including the nerves as 2D surfaces was chosen, and again, the absorption coecient of option of including the nerves as 2D surfaces was chosen, and again, the absorption coefficient of the the material assigned to the nerves was slightly increased to compensate for the surface lost and the material assigned to the nerves was slightly increased to compensate for the surface lost and the scattering coecient was estimated accordingly. Figure 5 includes a detail of one of the sides of the scattering coefficient was estimated accordingly. Figure 5 includes a detail of one of the sides of the octagon were the simplifications described above are shown, including the limestone canopies and the octagon were the simplifications described above are shown, including the limestone canopies and modelling of the limestone nerves considered in the dierent options. the modelling of the limestone nerves considered in the different options. Figure 5. Acoustic model of the York Minster ’s Chapter House. Detail of the simplifications including Figure 5. Acoustic model of the York Minster’s Chapter House. Detail of the simplifications including the limestone canopies and the three options tested for the limestone clusters of shafts or nerves framing the limestone canopies and the three options tested for the limestone clusters of shafts or nerves the stained glasses. framing the stained glasses. Acoustics 2020, 2 21 The final model built for the acoustic simulations, corresponding to option 2 in Figure 5, has 1927 planes and a total surface area of 2024.55 m . The characteristics of the finishing materials of the indoor surfaces of the room were determined by visual inspection and by consulting bibliography [16,29–34]. The Chapter House includes ceramic tiles for the floor, Magnesian limestone for walls, marble for the columns of the stalls, lath (strips of wood) and plaster for the vaulted ceiling, and last but not least, glass and lead for the stained glasses. The initial frequency-dependent values of the absorption coecients were assigned based on bibliography and publicly available databases [41,44,45]. The scattering properties were set to account for textures at mid frequencies, in the way that an estimation is made according to the depth of the decoration pattern of that surface as described in the Odeon manual [46]. The simulation model was adjusted through an iterative process in which on-site measurements and computer modelling results were compared [3,47]. The dierences found between them are assessed in terms of the just noticeable dierence (JND) values for various acoustic parameters [1,48]. The tuning process is based on the initial adjustment of the reverberation time by adapting the acoustic properties of the surfaces which present more uncertainties and which also have a significant surface area in the model, so that a relatively small change has a considerable eect. The absorption coecients initially assigned to those surfaces are carefully changed until the simulated values of the reverberation time at all frequency bands dier by no more than 5% (1 JND) from those measured on site. Once this primary requirement is achieved, a point-by-point comparison for other parameter values (EDT, T , C , D ) is performed considering the complete set of source-receiver (S-R) combinations. In those 80 50 S-R combinations where major dierences are found, a more detailed inspection of the RIR is done, analysing if there are any critical reflections added or missing in the simulation model. Having B-format or spatial RIR is crucial for this stage as it is possible to see where those reflections are coming from and implement local adjustments in the model if necessary. The model is considered adjusted once the dierences between measured and simulated values are acceptable for a plausible representation of the real sound field, meaning that the measured and simulated RIR are comparable in shape, and that the measured and simulated values of the acoustic parameters dier by no more than 2 JND (no more than 1 JND in the case of T ) in all the frequency bands [49]. A margin of 1 JND is desirable for all the acoustic parameters, but a margin of 2 JND is accepted taking into consideration all the approximations and limitations inherent to the measurement and simulation procedures. Furthermore, it is important to bear in mind that the calculation of the measured values through the use of dierent software tools may imply a variation up to 1 JND mainly due to the dierent algorithms they use to deal with the background noise and the arrival time detection [50]. Two dierent options have been tested for the adjustment of the acoustic model of the Chapter House. The first option considered the vault as the adjustment surface, since it is the finishing material that presents greater uncertainty in its acoustic behaviour. A relatively wide range of values have been published for plaster and lath works and other comparable materials [41,44,45]. The final value of the absorption coecient of the vault resulting from the iterative adjustment procedure increased from 0.16 up to 0.24 in the 125 Hz to 4 kHz frequency bands, which is greater than the range estimated by the dierent authors at mid-high frequencies (0.05 to 0.10). However, this value is in perfect harmony with the results derived from previous measurement campaigns developed in similar buildings [51]. Furthermore, considering that historical plasters have been shown to be more absorptive than modern ones [52] and that the vault has a large air cavity behind, the coecients obtained seem more than reasonable. The second option tested was to modify the acoustic properties of the limestone to achieve the adjustment of the model. The iterative process described above was applied, showing that the values of the absorption coecient initially assigned to the limestone at medium-high frequencies needed to be increased by 0.02 to match the measurements. On hard surfaces such as the limestone, such increment implies a relative change of almost 50%, which a priori seems large considering the polished finishing of the limestone surfaces found in the Chapter House, remarkable for the time of construction. Acoustics 2020, 2 22 Although the absorption coecients of the limestone resulting from the iterative process were high for a hard surface [41], similar values have been estimated for comparable materials in historical buildings before [12], and therefore, this option was not discarded. Furthermore, in order to achieve a good match in high frequency, an additional increase of the absorption coecient assigned to the stained glasses and the vault in the 2 kHz frequency band was required. Both options were based on reasonable assumptions, and therefore, tested in the simulation software. After testing both options, it was decided that the first option would be applied, as it resulted in a better reproduction of the early reflection pattern in terms of their intensity which led to a better match between measured and simulated acoustic parameter values. Table 3 summarises the acoustic properties assigned to the finishing surfaces of the simulation model of the York Minster ’s Chapter House, and the area which each of them occupies in the model. Table 3. Acoustic properties of the materials used in the simulation model of the York Minster ’s Chapter House, and the area which each of them occupies in the model. Absorption Coecients Scatt. Coef Area (m ) Frequency (Hz) 125 250 500 1 k 2 k 4 k 707 - Limestone walls and seats 0.02 0.02 0.03 0.04 0.05 0.05 0.05 686.8 Stained glasses 0.25 0.20 0.14 0.10 0.05 0.05 0.15 384.0 Lath and plaster vault 0.16 0.16 0.16 0.17 0.24 0.24 0.35 321.0 Tiled Floor 0.01 0.01 0.01 0.02 0.02 0.02 0.05 240.0 Limestone nerves 0.03 0.04 0.05 0.05 0.065 0.065 0.15 131.0 Limestone canopies 0.03 0.04 0.05 0.05 0.065 0.065 0.45 139.4 Marble columns 0.01 0.01 0.01 0.02 0.02 0.02 0.35 113.6 Wooden doors 0.14 0.10 0.06 0.08 0.10 0.10 0.20 8.8 1 2 Derived from tuning process; initial values raised up to compensate for the simplifications of the model. The final model is considered adjusted with the spatial averaged values of the acoustic parameters diering by no more than 1 JND except in the 250 Hz frequency band, where the simulated values are significantly overestimated, and in the case of EDT and T in the 4 kHz band, where the JND is much smaller in absolute values. Regarding the comparison point-by-point, more than 86% of the values match the criteria stated above (dierence between measured and simulated values below 2 JND) including all the frequency bands except 250 Hz. Figure 6 depicts the comparison of the spatially and spectrally averaged values (indicated by subscript “m”) obtained in each source–receiver combination. Following the indications of the ISO 3382-1 [1] T , EDT , T , D and C were calculated by averaging values obtained at 500 Hz 30m m Sm 50m 80m and 1 kHz frequency bands and J was calculated by averaging values obtained from 125 Hz to LFm 1 kHz frequency bands. The values corresponding to 1 and 2 JND have also been included in the figure to allow for the study of the margin of variation between the measured and simulated values of the acoustic parameters included in the comparison. Additionally, when comparing results, it is important to remember that the acoustic simulation is built under the premises of the geometrical acoustic theory, and therefore, can be considered reliable for frequencies higher than four times the Schroeder ’s frequency [41], which, in this case, is 64 Hz, meaning that simulated results in the 125 Hz and even in 250 Hz octave bands have to be been taken with caution. Acoustics 2020, 2 23 Acoustics 2020, 2 23 6.5 EDT (s) EDT (s) -- 6.0 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 5 5.5 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 4 -- -- -- 5.0 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 4.5 -- -- -- -- 2 4.0 3.5 125 250 500 1000 2000 4000 Frequency (Hz) 500 500 T (ms) T (ms) S Sm -- -- -- -- -- -- -- -- -- 400 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 350 -- -- -- -- -- -- -- -- -- -- -- -- -- 200 -- -- -- -- -- -- 300 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 100 -- -- -- -- -- -- -- -- 125 250 500 1000 2000 4000 Frequency (Hz) C (dB) C (dB) 80 80m -- -- -- -- -- -- -- -- -- -- -- -- -- -- -2 -- -- -- -- -- -- -- 0 -- -- -- -- -- -- -- -- -4 -- -2 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -4 -- -6 -- -- -- -- -- -- -- -- -- -- -- -- -- -6 -- -- -- -- -- -8 -- -- -- -- -8 -- -- -- -- -10 -10 -- -12 -12 125 250 500 1000 2000 4000 Frequency (Hz) 0.6 0.5 D (-) D (-) 50m 0.5 0.4 -- -- -- -- -- -- -- -- -- -- 0.4 -- -- -- -- -- 0.3 -- -- -- -- -- -- -- -- -- 0.3 -- -- -- -- -- -- -- -- -- -- -- -- -- -- 0.2 -- -- -- -- -- -- -- -- 0.2 -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- 0.1 0.1 -- -- -- -- -- -- -- -- -- -- -- 0.0 0.0 125 250 500 1000 2000 4000 Frequency (Hz) Measured values Simulated values 1 JND 2 JND Figure 6. Measured and simulated values of several acoustic parameters (from top to bottom: EDT, Figure 6. Measured and simulated values of several acoustic parameters (from top to bottom: EDT, T , TS, C80, and D50): Spectral behaviour of the spatially averaged values (left column) and spectrally C , and D ): Spectral behaviour of the spatially averaged values (left column) and spectrally averaged 80 50 averaged values obtained at each source-receiver combination (right column). The values values obtained at each source-receiver combination (right column). The values corresponding to 1 and corresponding to 1 and 2 JND have been also included. 2 JND have been also included. 3.2.2. Acoustic Simulation Details 3.2.2. Acoustic Simulation Details After the calibration process had been completed, the acoustic model was used to conduct a full After the calibration process had been completed, the acoustic model was used to conduct a full analysis of the current acoustics of the Chapter House focusing on the early reflection pattern and on analysis of the current acoustics of the Chapter House focusing on the early reflection pattern and on the behaviour of the space in the perimeter area of the room to evaluate the influence of its architectural the behaviour of the space in the perimeter area of the room to evaluate the influence of its features on sound behaviour. For this purpose, the initial range of source and receiver positions set architectural features on sound behaviour. For this purpose, the initial range of source and receiver during the measurement session was expanded, mainly to assess the acoustic behaviour of the space positions set during the measurement session was expanded, mainly to assess the acoustic behaviour in the area of the limestone stalls, used in the original configuration of the Chapter meetings, shown of the space in the area of the limestone stalls, used in the original configuration of the Chapter meetings, shown in Figure 3c. A new source position (S3) was placed in the seat reserved for the Dean (one of the central stalls in the East wall) to represent the position of a speaker and six additional S1_R01 S1_R02 S1_R03 S1_R04 S1_R05 S1_R06 S1_R07 S1_R08 S1_R09 S1_R10 S2_R01 S2_R04 S2_R05 S2_R06 S2_R08 S2_R09 S1_R01 S2_R10 S1_R02 S2_R11 S _R0 1 3 S1_R04 S1_R05 S1_R06 S1_R07 S 0 1_R 8 S1_R09 S1_R10 S2_R01 S2_R04 S2_R05 S2_R06 S2_R08 S2_R09 S1_R01 S2_R10 S1_R02 S2_R11 S1_R03 S1_R04 S1_R05 S1_R06 S1_R07 S1_R08 S1_R09 S1_R10 S2_R01 S2_R04 S2_R05 S2_R06 S2_R08 S2_R09 S1_R01 S2_R10 S1_R02 S2_R11 S1_R03 S1_R04 S1_R05 S1_R06 S1_R07 S1_R08 S1_R09 S1_R10 S2_R01 S2_R04 S2_R05 S2 06 _R S2_R08 S2_R09 S2_R10 S2_R11 Acoustics 2020, 2 24 in Figure 3c. A new source position (S3) was placed in the seat reserved for the Dean (one of the Acoustics 2020, 2 24 central stalls in the East wall) to represent the position of a speaker and six additional receiver positions (R12-R17) were set in various stone stalls (one in each of the sides of the octagon, excluding both the receiver positions (R12-R17) were set in various stone stalls (one in each of the sides of the octagon, East side on which the sources are located and the West side of the entrance door) as potential listener excluding both the East side on which the sources are located and the West side of the entrance door) positions (see Figure 7). as potential listener positions (see Figure 7). Figure 7. York Minster ’s Chapter House simulation model views. Figure 7. York Minster’s Chapter House simulation model views. The model was then modified to investigate the acoustics of the medieval Chapter House in its The model was then modified to investigate the acoustics of the medieval Chapter House in its original state. The aim of this simulation is to oer an insight into the probable acoustical impact that original state. The aim of this simulation is to offer an insight into the probable acoustical impact that the restoration works may have had on the space. It is always a challenge to try to plausibly reproduce the restoration works may have had on the space. It is always a challenge to try to plausibly reproduce the acoustic environment of a historic building at an earlier time, since there are usually gaps in the the acoustic environment of a historic building at an earlier time, since there are usually gaps in the information needed to create the acoustic model in terms of the finishing materials used and their information needed to create the acoustic model in terms of the finishing materials used and their acoustic behaviour. In this particular case, for example, there is no trace of the original floor of the acoustic behaviour. In this particular case, for example, there is no trace of the original floor of the Chapter House, and it is not certain how the original ceiling and the painted walls interacted with Chapter House, and it is not certain how the original ceiling and the painted walls interacted with sound, and therefore, some assumptions have to be made. sound, and therefore, some assumptions have to be made. The absorption coecient assigned to those limestone surfaces that were potentially painted The absorption coefficient assigned to those limestone surfaces that were potentially painted (see Section 2) has been reduced (0.01 at low frequency bands and 0.02 in mid-high frequency bands) (see section 2) has been reduced (0.01 at low frequency bands and 0.02 in mid-high frequency bands) taking into consideration how the paint finishing aects the acoustic characteristics of similar materials taking into consideration how the paint finishing affects the acoustic characteristics of similar according to the literature [44]. materials according to the literature [44]. For the 13th century wooden vault, the absorption coecient was estimated taking as reference the For the 13th century wooden vault, the absorption coefficient was estimated taking as reference published data on the acoustic behaviour of similar finishes and considering the resulting absorption the published data on the acoustic behaviour of similar finishes and considering the resulting coecient after the adjustment procedure in the current construction. Previous acoustic studies on absorption coefficient after the adjustment procedure in the current construction. Previous acoustic similar buildings revealed an averaged absorption coecient for wooden vaulted ceilings close to studies on similar buildings revealed an averaged absorption coefficient for wooden vaulted ceilings 0.40 [53]. Such absorption coecients were estimated from previous measurements campaigns in close to 0.40 [53]. Such absorption coefficients were estimated from previous measurements Italian churches [51]. Based on these results and considering the architectural characteristics of the campaigns in Italian churches [51]. Based on these results and considering the architectural original vault and its decoration at that time, an absorption coecient of 0.30 at low-mid frequencies and characteristics of the original vault and its decoration at that time, an absorption coefficient of 0.30 at 0.34 at high frequencies has been used in the simulations, which means an increment of approximately low-mid frequencies and 0.34 at high frequencies has been used in the simulations, which means an 10% with respect to the absorption coecients assigned to the current vault. increment of approximately 10% with respect to the absorption coefficients assigned to the current Finally, and even though the current tiled floor dates from the 19th century, no change in the vault. model has been introduced for this surface, since it is known that it was a similar hard surface Finally, and even though the current tiled floor dates from the 19th century, no change in the although presumably with a slightly more irregular and rough finishing, due to the time of construction. model has been introduced for this surface, since it is known that it was a similar hard surface Therefore, it is likely that this change did not have a major impact on the acoustics, since its absorption although presumably with a slightly more irregular and rough finishing, due to the time of and scattering coecients are unlikely to have changed significantly. construction. Therefore, it is likely that this change did not have a major impact on the acoustics, since its absorption and scattering coefficients are unlikely to have changed significantly. 4. Results and Discussion The results below are derived from the measured and the simulated B-format RIR, through the 4. Results and Discussion observation of the spectral and spatial behaviour of a series of acoustic parameters and the inspection The results below are derived from the measured and the simulated B-format RIR, through the of the reflection patterns found at each reception point. Results has been analysed and discussed in observation of the spectral and spatial behaviour of a series of acoustic parameters and the inspection of the reflection patterns found at each reception point. Results has been analysed and discussed in terms of perceived reverberation, clarity of sound and spatial impression, considering the different purposes this space has served through history. Acoustics 2020, 2 25 terms of perceived reverberation, clarity of sound and spatial impression, considering the dierent purposes this space has served through history. 4.1. Acoustics of the York Minster’s Chapter House Today The most relevant descriptor of the acoustic environment of a room is its reverberation time (T ). The analysis of the decay curves of the measured RIR shows an average value of T of 5.4 s. 30 30m As can be seen in Table 4, similar values are obtained in the lower frequency bands, in which the greatest values of standard deviation are observed, and T values significantly decrease at the higher frequency bands, where usually the sound absorption of materials is greater and the air absorption is more pronounced. The Chapter House has a large volume considering its total surface area and it lacks acoustically absorbent finishing materials, which leads to this high level of reverberation more typically found in churches or cathedrals with considerably greater volumes [54]. Its T is extremely high for a “meeting room”, since, typically, the recommended reverberation time to achieve adequate speech intelligibility in a conference or a meeting room of a comparable volume is around 1.1 seconds [55]. Its reverberation time is also above the values considered suitable for music reproduction (of around 2 s depending on the type of music), being even slightly above the limits of preferred values for organ music and medieval plainchant (set about 2–4 s) [56,57]. Table 4. Reverberation time (T ) and Early decay time (EDT) values measured under unoccupied conditions at the York Minster ’s Chapter House, together with their standard deviation (SD). Frequency Band (Hz) 125 250 500 1 k 2k 4 k T (s) 5.00 5.17 5.56 5.24 4.19 2.82 SD 0.13 0.13 0.06 0.05 0.03 0.02 T30 EDT (s) 4.97 4.94 5.45 5.04 3.92 2.36 SD 0.53 0.49 0.30 0.17 0.10 0.06 EDT spatially averaged including all the source-receiver combinations characterised. Looking at the Early decay time (EDT), which better assesses the subjective impression of reverberation [58], spatially averaged values are only slightly lower than those obtained for the T (Table 4). Nevertheless, SD values obtained for EDT are considerably higher than those obtained for T in all frequency bands, which indicates that the perceived reverberation depends on the relative position of the receiver in the room, meaning that the reduced source-receiver distance that certain source-receiver combinations have may emphasise the role of direct sound and the proximity to the surrounding walls may emphasise the contribution of early reflections arriving at certain receiving points. In any case, the room has an averaged EDT of 5.2 s, which is again above the optimal range suggested for speech (below 1.2 s) and for vocal and organ music in highly reverberant spaces, being 2.1 s and 4.2 s, respectively [57]. The analysis of the energy parameters derived from the measured RIR gives us complementary information about the balance between early and late reflections, or in other words, the clarity of sound in the space [58]. The definition, D (-), has been used to assess the clarity of the spoken word; whereas the clarity parameter, C (dB), was chosen to analyse the clarity of music. The central time, T (ms), 80 S which is strongly correlated with the decay time and therefore less sensitive to spatial variations [59], is useful to ascertain the clarity of sound in general terms. In Figure 8, the left column shows the spatially averaged values of the acoustic parameters measured for all the source-receiver combinations as a function of frequency. The high T mean values are indicators of a poor clarity of sound at low-mid frequency bands in the entire audience area. D mean values bellow 0.3 up to the 2 kHz band denote a poor clarity of the speech transmission in those bands, and the C values of less than 4 dB below 500 Hz are indicative of poor musical clarity especially at low frequencies. The error bars show the spatial dispersion in terms of the standard deviation (SD). It must be noted that they are relatively small, especially for the amphitheatre-like arrangement (S2), remaining mostly below 1 JND. Values of SD Acoustics 2020, 2 26 significantly exceeding this threshold are only found for the front-stage arrangement (S1) at the 500 Hz frequency band (max variation of 1.57 JND in the case of D at 500 Hz). The spectrally averaged values are also included in the figure as a function of the source-receiver distance. It can be seen that, when the source is located in the centre (S2) greater values of D and C are achieved, since the 50m 80m Acoustics 2020, 2 26 S-R distance remained below 7 m. As expected, there is a tendency for sound clarity indicators to decrease as the relative distance from the receiver to the sound source increases. However, it must be sound clarity indicators to decrease as the relative distance from the receiver to the sound source highlighted increases. Hthat owever, the r it eceivers must be high closest light to the ed tside hat th walls e receivers clo have higher sest to the sid C and e D walls h values ave high thaner those C80m 80m 50m that are at a similar relative S-R distance, but located in a more central area of the room. For instance, and D50m values than those that are at a similar relative S-R distance, but located in a more central D area of and the ro C om. values For inst obtained ance, D at 50m R4 and when C80m the va sour lues o ceb is tained located at in R4 when S1 (S1-R4 the dist source = 6.7 is m) locar ated e 1 in JND S1 50m 80m above of those obtained in R3 (S1-R3 dist = 6.4 m), which means that such dierence is perceptible (S1-R4 dist = 6.7 m) are 1 JND above of those obtained in R3 (S1-R3 dist = 6.4 m), which means that by such listeners. differen This ce is per fact may ceptible by be due listene to the r pr s. Th esence is fa of ct early may be du reflections e to the near pr er es to enthe ce of ea walls. rly In re any flecti case, ons the values of the energy parameters denote poor clarity of the sound (D < 0.3 and C <