arXiv:1905.13578v1 [physics.pop-ph] 31 May 2019 anrt aruhfripeetn i etvlidea festival his Richard implementing attracted for that house Bayreuth municipal a opera to as this Wagner used was often It more theatre. and more opera protec- the became as fire house installed a subsequently been and have lightening system the 45%. tion systems, roughly of by heating depth reduced air The was Warm performance place. for area took stage renovation 1817, had until first intact house largely the kept opera when hence Margravial and used the death been 1758, the rarely After in mu- cost. Wilhemine Wilhemine, maintaining composing of high by as the well used of despite as regularly sic, librettos been writing enjoyed had of who it daughter only 1748. then, originally in the the Since was Wilhelmine of in Margravine house marriage and house Friedrich opera the Margrave opera court celebrate emphasis the This to an to built adapted years. with made 270 House [3], changes past Rainer Opera main by the the edited of on book history the short from a is Below the This after year a acoustics. renovation. half its recent roughly to of conducted house is evaluation opera measurement this objective of re- an response a provide [2], impulse Festspielhaus following the Bayreuth measure fact, we the this of by far characterization is Motivated cent house opera discussed. this of less acoustics the 18th photographs, tours and the site through with perceivable readily fa¸cade along are Baroque acous- decorations the and interior While form [1] loge extraordinary properties.” tiered iconological and its decorative of tic, Giuseppe terms by in architecture Bibiena theatre Galli court master- Baroque a is of House work Opera outstanding Margravial its “The for that was justification value one nat- 2012, and in cultural world heritage being of ural list was UNESCO’s Bayreuth the in House inscribed Opera Margravial the When Introduction similar a of theatres Italian historical age. to discussed time and comparison decay characterized in are reno- early factor (RT), recent clarity its and time (EDT) after reverberation house The opera vation. this im- of the characterize response we pulse sources, balloons sound cen- Using dec- as hand-claps 18th the intact. and with remains the canvas along in painted auditorium theatres oratively also bell-shaped but opera the as visual of tury, an the design provides only acoustic It not the experience [1]. Bibiena to theatre Galli opportunity court Baroque Giuseppe preserved between by well designed a built is Bayreuth, 1750, and House 1745 Opera Margravial The Abstract eata rus ienVokl hitp onr lxnr V¨ Alexandra Dobner, V¨olkel, Christoph Simeon Krauss, Sebastian xeietlhskV nvri¨tByet,940Bayre 95440 Universit¨at Bayreuth, V, Experimentalphysik cutc fMrrva pr os Bayreuth House Opera Margravial of Acoustics 1 mi:[email protected] Email: 2 et itiue nteflo,treteso oe,the gallery. loges, of the tiers and three loge floor, court the on distributed seats 520 h uioimhsattlvlm faot5000m about of volume canvas. total with a of covered has Part are auditorium ceiling parts. The (abies) the other and fir the walls while for wooden used the sculptures, were lin- the (picea) specifically, for spruce More and used original was [3]. its (tilia) wood retains den i.e., auditorium [11]. the material, architects floor, building the the of acous- for belief the Except the improving the for representing for hand inside, hand other tics the one on the and on dec- festival, heavily designed The in are [11]. ornaments house the theatre orative resembles “ideal” design opera Lamberti’s This of loges. Margravial plan of a tiers the stalls, three including and auditorium 1, gallery shaped bell Fig. a has Bayreuth in shown As standard a ISO [9]. following 3382 provides characterizations acoustic which room for [8], routine ITA-Toolbox Matlab with the The analyzed using consequently respectively. are devices, signals as recording recorded used and also are sources HCSP, microphones to sound condenser addition and in Here, balloons for the air [2]. except characterize regime well, frequency reasonably to low room the used a be of time can reverberation record- approach phone was (HCSP) smart it investigation, ing and previous hand-claps a that In demonstrated 7]. [6, phones be- smart are in microphones embedded systems) characterizations quanti- and (micro-electro-mechanical such claps subsequently MEMS hand Nowadays, using be convenient more can coming 5]. room [4, a fied of and on features reverberation, music based other echoes, strength, intelligibility, response, the speech on impulse clarity, characterizations opera the further Margravial measuring which the by of acoustics house the characterize We Procedure Measuring century. very 18th be the in should status nowadays original house the its to area, opera close proscenium the set the of stage to acoustics the modifications floor, possible wooden and non-original the summa- To beside had layer strengthened. rize, painting and original stabilized the recovered, of been opening. 90% roughly stage original which dur- the of (2013-2018), ing renovation restore recent size most to the reduced reversed in substance a been of had front as changes in well Those stairs as new stages, balustrade, wooden the curving the a of with replacement railing the including had opera the modified, area During this proscenium been was. the that it (1935-1936), as renovation considered exactly second kept later be better he should house although 1871, in t,Germany uth, le,adKiHuang Kai and olkel, 1 3 with abled. In addition, three condenser microphones (one Superlux ECM999 and two AKG C1000S) are also used as recording devices. Except for the combination of SS4 with M1, the distance between each sound source and recorder is at least twice the reverberation distance [5]. The sampling rate for all recordings is 44.1 kHz. The recorded signals from all microphones are automatically cut into individual impulses. Using the ITA-tool box, y the raw signals are filtered into different octave bands and subsequently the energy decay curves for each band (EDC) are obtained through applying the backward in- tegration method on the squared envelope of the filtered X SS3 Stage X X signals [4]. From the least square fits of individual EDCs SS2 X X X X X in a semi-logarithmic plane, the reverberation times are x SS1 backdrop obtained. Depending on the different ranges of data used in the fitting, EDT (early reverberation time), T15, T20, T30, etc., are obtained. The uncertainty for the RT is es- timated from the residuals around the regression line [12], taking into account their autocorrelation [13, 14] arising Figure 1: Upper panel: The interior of the Margravial opera from the backward integration method. The autocorre- house viewed from the court loge [10]. Lower panel: Ground lation is approximated by its upper bound 1 to have a plan of the Margravial Opera House Bayreuth adapted from conservative error estimation. In the present study, EDT, Ref. [3] with the borders of the auditorium, stage, as well as RT and clarity index C80 are obtained in the frequency the locations of the sound sources (SS) and recording micro- range from 125 Hz to 4000 Hz, covering six octave bands. phones (M) marked. M8 is located in the court loge opposite to the stage. The gray shaded region marks orchestra area. See Table 1 for measured coordinates in the Cartesian system Results and Discussion defined by dashed arrow lines. The thick red line corresponds to a iron curtain, which was closed for measurement SS4 and open for SS1-3. 1 0.2 0.5 a) 0.1 0 0 Table 1: Positions (in meter) of sound sources and recording (arb. unit) -0.5 -0.1 (arb. unit) devices in the coordination system defined in Fig. 1(b). Balloon Clap

audi. -1 -0.2 audi. SS1 SS2 SS3 SS4 p 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 p X -17.7 -2.7 -2.7 4.1 Time (s) Y 0 0 2.4 0 0 b) M1 M2 M3 M4 M5 M6 M7 M8 -10 X 6.5 10.7 14.9 19.1 7.8 12.3 17.5 23.5 Y 0 0 0 0 4.1 3.3 2.4 0 -20 -30

The measurement was conducted on October 11, 2018. 0 In the auditorium, the temperature and relative humidity c) in the auditorium was 18.9 degrees and 45%, respectively. -10 Impulse signals are generated either by air balloons (AB) -20 or hand-claps (HC). The balloons are inflated with a ± EDC Balloon EDT pump to a diameter of 23 3cm (measured at the equa- Energy decay curve, EDC (dB) -30 tor) and pricked with a needle (outer and inner diameter EDC Clap T20 2mm and 1.4 mm, respectively) to generate impulse sig- 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 nals. Both AB and HC signals are released from a height Time (s) of ∼1.8 m above the ground. Two room conditions with four sound source (SS) locations have been chosen: Iron Figure 2: (a) Representative raw signals from one balloon- curtain closed (SS4) or open (SS1-3). At each SS loca- burst and one hand-clap at SS3 recorded by M3. (b, c) Corre- tion, 10 AB followed by 10 HC signals are generated and sponding energy decay curves of the impulse from the bursting subsequently recorded by the microphones distributed in balloon and from the hand-clap at the 1000 Hz octave band. the auditorium. Six smart-phones are clamped to the The orange and green solid lines correspond to linear fits for chairs with flexible arms and oriented to face the stage. obtaining the early decay time and T20, respectively. For the The levels of the microphone gain are adjusted before EDC curves, only every 600th data point is shown for a better each measurement and the automatic gain control is dis- visibility. Figure 2 shows a comparison of two sample raw signals visibility, only T20 values are shown here as representa- generated by a balloon-burst as well as by a hand-clap. tives. The other RT values agree with T20 within exper- It shows that the sound pressure level (SPL) generated imental uncertainties, provided that the fitting criterion by hand-claps is roughly one order of magnitude smaller described above is applied. The reverberation times of than that from balloon-bursts. Consequently, the EDC the auditorium for low (125 and 250 Hz octave bands) curve in the former case decays rather quickly to the and middle (500 and 1000 Hz octave bands) frequency surrounding noise level in comparison to the latter case. ranges, averaged over various sound sources and micro- Therefore, the characterization of RT with T20 or above phones are 1.67 ± 0.11s and 1.29 ± 0.02 s, respectively. becomes unfeasible, as the fitted curve for T20 in (c) This results lead to a bass ratio (BR) of 1.29, matching shows. In order to obtain reliable RTs, we set a thresh- the grand average (1.30 in unoccupied conditions) of 50 old for the uncertainty obtained from a least square fit: traditional Italian opera houses characterized recently by Only RTs with a relative uncertainty smaller than 10% Prodi et al. [16]. are used in the following analysis. With this criterion, un- When the iron curtain is closed (condition for measure- realistic RTs [e.g. fit in (c) delivers T20=5.2 ± 1.9s] can ments taken at SS4), the overall reverberation time re- be sorted out. Only data obtained from balloon experi- duces to 1.29±0.05sand 1.28±0.03 s for low- and middle- ments are used in the following characterization, because frequencies, respectively. According to the Sabine for- the impulses generated from bursting balloons are more mula [17, 4], RT ∝ V/S with total volume V and surface reliable concerning repeatability [15] and feasibility. area S of the room. As V is dramatically reduced with closed iron curtain, and in the meanwhile S does not change with the same proportionality as V (due to the decorative, sound absorbing surfaces), the overall RT is expected to decrease substantially. The other obvious feature is the vanishing spectral variation, which arises presumably from the vanishing coupling between the au- 2 SS4 ditorium and the stage [4]. SS1 1.8 (s)

20 Table 2: Averaged reverberation time obtained without 1.6 (SS1) and with (SS4) ironcurtain closed, unoccupied.

1.4 125 250 500 1000 2000 4000 Hz SS1 1.79 1.61 1.37 1.24 1.18 1.11 s 1.2 SS4 1.25 1.33 1.31 1.25 1.25 1.15 s Reverberation time T 1 To estimate the transparency of music in this opera 125 250 500 1000 2000 4000 house, we also characterize the clarity index C80 [4], Frequency (Hz) which characterizes the relative importance of the early part (within 80ms) with respect to the later part of the impulse response, and hence provides information on the effectiveness of sound reflections. For all the combina- tions of SS and M, C80 ranges from 0.5 to 3.2. The average value of C80 in the stalls is 1.9 ± 0.6 and falls in the preferred range by Hidaka and Beranek [18]. It Figure 3: Reverberation time (T20) of the auditorium at suggests that the subjective clarity is sufficient even for different octave bands for two different conditions: With the fast musical passages. Together with the averaged EDT iron curtain open (SS1) or closed (SS4). It is averaged over ∼ 1.12 ± 0.15s (with iron curtain open), we can com- all microphones except for M2, which has technical issues. pare the Margravial opera house with traditional Italian The error bars correspond to the maximum of the uncertainty theatres as well as modern theatres after the work of from fitting and that from data scattering among different Prodi et al. [16]. It is interesting to see that the Mar- microphones. gravial opera house falls into the group of “regular” the- atres without large hard reflecting surfaces. Although The spectral variation of the reverberation time is shown C80 has a clear spatial distribution in the auditorium, in Fig. 3 and listed in Tab. 2. Similar to ‘regular’ Ital- the relatively small standard deviation 0.6 suggests that ian historical opera houses, there exists a clear increase the overall good transparency and clarity for music, at of RT as frequency decreases to the 125 and 250 Hz oc- least in the stalls that were characterized. tave bands, arising from the extensive mid- and high- frequency sound absorptions due to the interior de- Conclusions and Outlook sign [16]. For instance, the loges act as resonators for low frequency sound, and meanwhile as effective absorbers To summarize, room acoustic characterizations are con- for mid- and high-frequency sound, owing to the deco- ducted for the newly renovated Margravial opera house rative design and sound absorbing finishes. For a better in Bayreuth. Our investigation suggests that the acous- tics of this opera house matches perfectly that of typical mobile app zur bestimmung raumakustischer pa- “Italian-style historical opera houses” built between 1637 rameter. In the 43th German Annual Conference and 1887 [16], representing the belief of the architects on on Acoustics (DAGA), 2017. “ideal” theatres [11]. It has a relatively low reverbera- [8] Pascal Dietrich, Martin Guski, Johannes Klein, tion time in comparison to modern theatres due to the Markus M¨uller-Trapet, Martin Pollow, Roman decorative interior design and the lack of large hard re- Scharrer, and Michael Vorl¨ander. Measurements and flecting surfaces. It delivers fairly well clarity at least for room acoustic analysis with the ita-toolbox for mat- the stalls. For the loges and boxes, C80 is expected to lab. In 40th Italian (AIA) Annual Conference on be influence by detailed configurations of seating in the Acoustics and the 39th German Annual Conference boxes. Therefore, further characterizations in the loges, on Acoustics (DAGA), 2013. particularly with the presence of audience are needed to understand its acoustic design better. [9] ISO3382-1, acoustics measurement of room acoustic parameters. part 1: Performance spaces, 2009. From a technical perspective, this investigation shows that bursting balloons serve as a more repeatable handy [10] http://www.bayreuth-wilhelmine.de [accessed 2019- sound source in comparison to hand-claps. For recorded 02-01]. signals with relatively weak signal-to-noise ratio, setting [11] Patrizio Barbieri. The acoustics of italian opera up a threshold for the goodness of fit is valuable for houses and auditoriums (ca. 1450-1900). Recercare, obtaining more reliable and reproducible reverberation 10:263–328, 1998. times. [12] J. Higbie. Uncertainty in a least-squares fit. Amer- Acknowledgments ican Journal of Physics, 46(9):945–945, September 1978. We thank Thomas Rainer from the ‘Bayerische Schl¨osserverwaltung’ and Angela Danner from the press [13] Cathy Campbell. Properties of ordinary and office of Bayreuth University for bringing us the oppor- weighted least square estimators of regression coef- tunity of conducting the measurement. The detailed in- ficients for two-stage samples. In Proceedings of the formation on the history and technical aspects of the Social Statistics Section, American Statistical Asso- opera house kindly provided by Thomas Rainer helped ciation, pages 800–805, 1977. greatly in conceiving the original acoustic design of the [14] Brent R. Moulton. Random group effects and the theatre. SK acknowledges support by the Elite Network precision of regression estimates. Journal of Econo- of Bavaria (Study Program Biological Physics). We are metrics, 32(3):385–397, August 1986. also grateful to Tobias Eckert, Christoph Schnupfhagen, Nico Stuhlm¨uller, and Michael Seidel for their kind help [15] Jukka P¨atynen, Brian F.G. Katz, and Tapio Lokki. in audio recording. Investigations on the balloon as an impulse source. The Journal of the Acoustical Society of America, 129(1):EL27–EL33, January 2011. References [16] Nicola Prodi, Roberto Pompoli, Francesco Martel- [1] UNESCO World Heritage Centre lotta, and Shin-ichi Sato. Acoustics of Italian His- https://whc.unesco.org/en/list/1379/ [accessed torical Opera Houses. The Journal of the Acoustical 2019-02-01]. Society of America, 138(2):769–781, August 2015. [2] K. Huang. Impulse response of the bayreuth fest- [17] W. C. Sabine. Collected papers on acoustics. Har- spielhaus. In Fortschritte der Akustik, pages 238– vard University Press, 1922. 241, Kiel, Germany, 2017. [18] Takayuki Hidaka and Leo L. Beranek. Objective and [3] Thomas Rainer. Margravial Opera House Bayreuth. subjective evaluations of twenty-three opera houses Bayerische Schloesserverwaltung, 2018. in Europe, Japan, and the Americas. The Journal of [4] Heinrich Kuttruff. Room Acoustics. CRC Press, the Acoustical Society of America, 107(1):368–383, London, England ; New York, NY, 4th edition edi- December 2000. tion, October 2000. [5] Gerhard M¨uller and Michael M¨oser. Handbook of Engineering Acoustics. Springer, Berlin, Heidelberg, November 2012. [6] Prem Seetharaman and Stephen P. Tarzia. The hand clap as an impulse source for measuring room acous- tics. In Audio Engineering Society Convention 132, Apr 2012. [7] Andreas Rosenkranz, Ralf Burgmayer, David Acker- mann, Markus H¨adrich, and Stefan Weinzierl. Eine