On best acoustical parameters’ values for ‘ Liederistic ’ music performance: A preliminary study

Abstract

‘Liederistic’ music is a musical genre on which architectural acoustical research is rarely focused. Despite this lack of scholarship, however, this genre represents one of the most interesting challenges for acoustical problem solving. In fact, it combines speech and music comprehension requirements in a fascinating manner to illuminate elements of both the magnification of the musical score and the hidden tension of a sung poetry text. To individualise the main characteristics of this genre, a multidisciplinary approach is taken, starting from a musical point of view. A knowledge and comprehension of music is a fundamental step to determine the parameters that allow for the best performance and listening experience of this particular music. Consequently, an introduction to the physics and acoustical characteristics of the singing voice and the piano is given, specifically highlighting the technological development of the piano since the 18th century. Through a combination of historical, architectural, and interior design research, a possible location of the original performance is recreated to investigate the probable primal acoustic conditions using computational analysis. Various contemporary chamber music halls were then considered and compared to the reconstructed location to investigate whether actual places exist with similar acoustic conditions.

Keywords

Architectural acoustics Lieder chamber music Schubert singing voice piano acoustic reconstruction musical acoustics

Introduction

Acoustical phenomena have fascinated human beings since ancient times. It was often interpreted on the borderline between scientific manifestation and magic; through this vision, men of culture played an important role in spreading knowledge of this incredible subject. One of the most famous of such characters was the Jesuit Athanasius Kircher, who understood the dichotomy of sound from both magical and scientific points of view. He studied and experimented ancient legends of talking statues while at the same time making important observations on anomalous sound behaviours in 17th-century buildings, such as Villa Simonetta in Milan.¹ The relationship between sound and buildings is an ancient matter, as we have recorded efforts pertaining to sound control from the ancient Greek ruins,² but a purely mathematical and physical approach is quite recent. The American physician Wallace Clement Sabine was the first to rationalise the length of a sound’s perception in his famous formula in the first years of the 20th century, which he called reverberation time (RT).³ Since then, many improvements have been made by physicists to increase the number of tools we can use to control sound in an enclosed space. Despite such efforts, however, establishing parameters to control sound is still a current topic of discussion. For example, the reverberation time’s direct relationship with the architecture of the space, represented by its volume, was one of the main tools used to elaborate the preliminary part of an architectural acoustic project. Studies demonstrate that inside historical buildings like Romanesque churches, there is not a significant correlation between RT and the space’s architecture because churches may have the same reverberation despite differences in their volume distribution and architectural configuration.⁴ Different situations, like the design projects for an orchestra rehearsal room, show the importance of preliminary investigations regarding sound waves and sound levels for performers and their relationships with an average volume for each musician, outlining a sort of life-perimeter.⁵ There is a correlation between sound and musicians’ space and the suggested design method starts from the optimal conditions for a rehearsal performance to realize the hall which allows them. Following this method, starting from the acoustical phenomena to reach the structure and the acoustic of the architecture, this paper attempts to summarise and determine the correct approach and parameters for realising the most suitable acoustical performance for a particular genre of music: Lieder (art songs). These kinds of ‘songs’, which were popular during the 19th and 20th centuries and are often associated with the Austrian composer Franz Schubert, heavily rely on some of the tools we used during sound control projects. This paper represents a basis for analysing this sophisticated and not commonly scientifically-analysed genre from an architectural acoustician’s point of view. A multidisciplinary approach was assumed; the article begins with a description of Lieder’s musical features and characteristics and then explores social and historic facts concerning the genre, as well as the acoustical properties of the instrument and performers. Finally, a possible original stage of this genre’s performances is created, and its acoustical properties are compared to the main contemporary chamber music hall.

State of the art

Schubert’s Lieder

Many great composers of the past used to write chamber music. Sometimes it could be an exercise, an experiment in musical harmony or instrumental musical-line control; other times it represented a way to discover new sound expressions. Since the late 17thcentury, canonic composers such as J.S. Bach wrote music for small instrumental ensembles. From the second part of the 18th century, famous composers such as W. A. Mozart, J. Haydn and L. Van Beethoven wrote in the so-called ‘chamber music’ genre, the most famous manifestation of which was the Quartetto formation, usually comprised of two violins, one viola and a cello. Chamber music’s golden period was during the late 18th and the early 19th centuries and it represented an expression of the meetings among the upper middle class. During this time, successful composers such as Robert Schumann played many pieces for piano and voice. This particular kind of piece is referred to as a Lied (plural Lieder), the German term for ‘art song’. A ‘Liederistic’ production may be one the most famous genuine genres of the early Romantic (or late-Classical) period and it is largely associated with the Austrian composer Franz Schubert (Figure 1(a)). He was a prolific musician based in Vienna during the early years of the 19th century, and appreciated by Beethoven himself. He died at the age of 31 in 1828. He composed more than 18 vocal masses, 33 orchestral works, 36 piano sonatas, 800 vocal works and 21 quartets.

Figure 1.

(a) Oil portrait painting of the young Schubert by Wilhelm August Rieder, after his own 1825 watercolour portrait, Vienna Museum, 1875. (b) and (c) are extracts from Schubert’s Lieder. In (b) (measures 1–8 ‘Am See’ D.746), the piano plays an important role, reflecting the continuum movement of water Its dynamic shows a shining anxiety as represented by the dominant seventh harmony of the second part of each measure that influences the voice score even in the notes’ choice. In (c) (bars 5–8 ‘Die Forelle’ D.550), the piano creates all the ‘movement’ of the scene, with the chord jumps in the left hand and with the semiquaver movement of the right hand that collides in the accent in the second part of the bar. The accented note can be analysed with the voice’s note, creating an interesting tension of the dominant seventh chord in bar 7.

His vocal work series included 600 pieces for vocal soloist and piano accompaniment. Schubert wrote these works for any kind of vocal extension and the most famous series is the song cycle, whose songs are based on poems that tell stories and legends using both imaginary and realistic characters. Different kinds of studies have been conducted about this fascinating genre of music, discovering various aspects that were hidden from a superficial analysis. The most important change Schubert made to the Lied was rebalancing the poetry text and musical accompaniment. He enhanced the role to the piano score (Figure 1(b) and (c)), which was a revelatory decision. In fact, despite assumptions regarding the terminology, the accompaniment is not enriched until its role equals that of the voice; the aim of the song was not primarily to express feelings or meanings first located in poetry, but instead ‘to align music with the widespread effort of literary and philosophical Romanticism to represent subjectivity in action’.⁶ For this reason, Franz Schubert was labelled ‘the balancer’ or, using the words of Schleiermacher,⁷ the composer inspires the ‘Divination’ because the often independent role of the piano and its complex interaction with the voice results in continual friction between the apparent primacy of what the poem says and the ascendancy of what the music divines. This friction replaces the unauthoritative vocal subject with the so-called emotional sublime – the claim of fellow feeling serving to justify the passions of the alienated and abject figures.⁸ In other words, the ‘I’ of the protagonist becomes thinner and a pervasive call to sympathise with troubled figures like the wounded is manifested. A thorough analysis has been made regarding the pathological protagonists of the song cycles, ‘Die schone Mullerin’, the ‘Winterreise’ and the ‘Schwanengesang’, revealing their complexity.⁹ Stemming from these analyses, it is logical to assume that the voice and piano require the same importance and visibility in their acoustical treatment.

A historical approach to the acoustics of Schubert’s time

The parameters and relationships that are currently used in architectural acoustics were developed more recently than both this genre of music and the construction of the places where Lieder used to be played. For this reason, taking a historical approach to find the main acoustical properties of the performing space should prove beneficial. In fact, it is well known that many composers adapted their compositions according to the acoustics of the place where they were to be played, a phenomenon exemplified by Joseph Haydn and many of his works. The rooms where his pieces were played varied from 1530 and 6800 $m^{3}$ , with calculated reverberation times of about 1.0 to 1.7 s. The smallest hall was the Musiksaal of the Esterhàza Castle, and the biggest one was the Haydnsaal of the Eisenstadt Castle. How Haydn used the acoustics of these halls for his compositions are fairly demonstrable and many examples have been identified.¹⁰ For example, the dynamics and orchestration of the beginning of the 61st Symphony, a fast passage that begins with the full orchestra at a forte (loud) dynamic level to the first violin section at a piano (soft) level, requires a very dry acoustic. On the other hand, a reverberant acoustic is suitable for other (and later) pieces like the 102nd Symphony, where, after the forte chord of the full orchestra, there is a measure of rest to allow the sound to extinguish before restarting.¹¹ Another important example is the long pause in the famous Toccata and Fugue in D minor by J.S. Bach, where the composer took advantage of the long reverberation of a church to better highlight the evocative introduction of the initial trill.¹² Other examples include organistic productions by César Frank during the 19th century, where the pauses and stop lists are chosen according to the velocity and dynamic of the piece, or the metronomic attention in symphonic productions given by the later organist Charles Marie Widor, as seen in the first page of his sixth symphony.¹³ Such examples are sufficient to indicate that composers consider the acoustics of the place where their pieces were played. In the case of Franz Schubert, the performances of his Lieder took place in the magnificent rooms of the palaces that belonged to the aristocracy of the period. Thanks to the biographical work completed by Gibbs¹⁴ and the available data on the Schubert legacy,¹⁵ it is possible to summarise the halls where Schubert frequently played:

Franz Von Schober’s palace (the first Schubertiade, 21 January 1821)

Atzenbrugg Castle (1821–1822)

Baron Josef von Spaun apartments in Linz

Sonneleithner apartments in Wien

Salon Hofrat of Josef Witticzek

Johann Heinrich Geymuller’s Palace

Salon of Sylvester Paumgartner in Steyr (Currently a Reno shoe store)

Salon of Marie Pachler, also known as the House Zum Rabenschinder in Herrengasse, Graz (unfortunately not the original building because it was rebuilt in 1890)

Unfortunately, no acoustics data have been recorded for these halls, as the majority are either not in their original configuration or no longer exist, so it is difficult to forecast any site instrumental analysis. One possible solution, however, is to analyse the most common places where music used to be performed in Vienna during the early 19th century. There were three theatres in Vienna: the Theatre ‘An Der Wien’, the Theatre ‘Mehlgrube’ and the Hofburg Theatre Palace, currently known as the ‘Spanische Hofreitschule’.¹⁶ The only one that still exists in a similar condition and configuration to Schubert’s life is the ‘An Der Wien’ Theatre. Meyer¹¹ made an acoustics analysis and reported a reverberation time of 1.15 s at mid-frequencies. As one of the more popular music performance venues, Schubert must have been familiar with it and its acoustics. This data could be a source of investigation for determining or assuming the reverberation time in which the composer worked. Unfortunately, no other data are available at this time.

The piano technology in Schubert’s time

It is important to talk about the mechanics and style of pianos that Schubert used. During the early years of the 19th century, the piano was a relatively young instrument, differing significantly from the modern Grand piano we now see in concert halls. The American pianist Bilson¹⁷ wrote an interesting paper on this subject in 1980. He first explained the differences between the most popular key transmission technologies and how each was developed from the late 18th century to the first decades of the 20th century. He then integrated his study with his personal artistic experience as a pianist and performer of 19th century music. The dissertation was written to demonstrate how the ‘sound’ that Schubert required from his piano pieces is softer than we assume from our modern point of view because of the key transmission of the type of piano he likely played on. The instrument most likely had a Wienermechanik (Viennese action) and consisted of the following arrangements¹⁸:

The key is directly linked to the hammer by a Kapsel (fork);

The wood hammer is softly covered by multiple leather layers; and

The tension of the strings is very moderate, which allows a wooden frame to carry it.

These characteristics are different from the harder hammers and greater string tension carried by a metal frame and the English action that other piano producers (Pleyel, Erder, Steinway, Bechstein, and Bosendorfer) required to make louder pianos in the increasingly larger concert and recital halls of the time. The advantages of Viennese action include its ability to facilitate greater agility and produce unique piano dynamics. With these instruments, it is possible to play a real ppp (a very soft pianissimo) without a virtuosic and technical approach to the keyboard. Thanks to this aspect and other important albeit strictly musical reasons (that are omitted in this technical acoustical dissertation), Bilson assures readers that Schubert’s piano music is meant to sound softer than we might assume. This makes the piano accompaniment’s interaction with the voice part even more interesting.

The piano and the singing voice

The previous paragraphs demonstrated that the process for recovering the original acoustic conditions of Schubert’s lifetime is rather difficult. Through an acoustical introduction to the physics of the main performers of the art song, however, differences between a contemporary performance and the original one can be derived. The following reported measurements were carried out in an anechoic chamber with a microphone distance of 3.5 m.

The piano

Because the piano is a very complex percussive musical instrument, a brief description is necessary to highlight the differences between the modern-day Grand piano and the Viennese action piano; however, a comprehensive physical description of the instrument goes beyond the intention of this paper. For this reason, the loudness and directionality of the instrument will be discussed, while the so-called Time structure will be disregarded at this time.

Loudness

The piano’s soundboard has a key role in determining the sound spectra of the instrument due to its resonance, which is typically defined within the 200–1000 Hz interval range (but it has been extended to 100 Hz by increasing the dimensions).¹¹ This characteristic causes an unequal sound pressure level; the level drops 10 dB per octave when the frequency is equal to or above 1500 Hz.¹⁹ However, the sound pressure level of the piano is mainly determined by the dynamic choices of the pianist during the key attack. Table 1 summarises the most simple and common situations.¹¹

Table 1.

Grand Piano’s sound pressure levels depending on the dynamic, piano-lid configuration and number of contemporary played notes.

Action – two voices scale (playing two notes at a time)	Dynamic	Sound pressure level (dB)
Open Lid	ff	104 (106 in low registers)
Open Lid and right pedal	ff	107 (108 in low registers)
Closed Lid	ff	102
Open Lid	pp	88
Open Lid and left pedal	pp	87
Closed Lid	pp	86
Closed Lid and left pedal	pp	85
Playing a single note (Open Lid)	pp	65

Directionality

Although the piano has an unequal sound pressure distribution, what most accurately characterises the directionality of the sound produced is the lid configuration. In fact, although the resonant behaviour of the soundboard determines the sound distribution, the four possible opening positions of the lid (open, closed, half open, and removed) affect the final result, shadowing the sound (Table 2).¹¹

Table 2.

Effect in the register of a Gran piano depending on the lid configuration.

Configuration	Effect in the register
Open Lid	In the low register, there is mostly uniform sound radiation; the soundboard behaves as a dipole at low frequencies (250 Hz), which generates a difference of 5 dB between the back of the piano and the front lid. This also creates the illusion that the sound source is above the piano at a certain distance.
	In the middle register, the lid starts to shadow the sound and the major pressure level of the higher frequencies (i.e. 500 Hz) starts to be located to the right of the performer.
	In the high register, the lid characterises the sound with evident affection (i.e. at 4000 Hz there is a sound pressure difference about 12 dB between the right and the left of the pianist).
Closed Lid	In all the registers, a great uniformity is recorded between the right and the left of the pianist.
Half Open Lid	The 10° lid opening caused a 6 dB decrease near 4000 Hz but has a very similar profile to the open lid configuration.
Removed Lid	Without the lid, the piano is equalised in all the directions, with a decrease only on the lower register (below 250 Hz), localised close to the horizontal plane (extreme 0° and 180° of the pianist vision).

Therefore, sound emission and directionality highlight two important aspects of the present analysis:

The metal soundboard determines the sound power and diffusion. For this reason, future studies focused on the Viennese action piano will be fundamental to understanding how a wooden soundboard affects the resonance and directionality of the instrument.

The equal distribution of the sound that allows the closed lid configuration can be important in understanding why the piano in the paintings (see next section) is always in the centre of the hall with the closed lid during a performance.

The singing voice

Unlike a piano, the human voice does not depend on the soundboard’s material resonance but rather on the sung vowel. In fact, the sound’s spectra are mainly characterised by formants, and the strongest partial’s location changes depending on the vowel emitted by the singer. For example, a male voice should have a low formant located between 150–900 Hz and a second one between 500–3000 Hz, but it drops off 15–20 dB at frequencies below the first formant.¹¹ Furthermore, the sounds change depending on the sung text and the vowel location inside each word, thus extending the possibility of the emitted spectra to semi-infinite possibilities. Interesting studies led by Winckel²⁰ and Sundberg²¹ in the 1970s showed how the human singing voice and the singer’s formant allows the performer to transcend over an orchestra due to the weaker overtone diffusion of its musical instruments.

As with a piano, it is possible to analyse the human singing voice by sound power and directionality.

Sound power

Although formant and vowels determine the pressure level of the singer’s voice, the performer’s dynamic choices mainly determine the loudness of the emitted sound. Table 3 presumes the singing voice dynamics as catalogued by Burghauser and Spelda.²²

Table 3.

The lowest, highest and overall sound pressure levels of a singer depending on the voice type.

Solo voice	Sound pressure level extension in the lowest register (dB)	Sound pressure level extension in the highest register (dB)	Overall sound pressure level extension (dB)
Bass	65–82 (A1)	102–108 (A4)	62–112
Baritone	68–80 (F2)	82–102 (G4)	65–106
Dramatic tenor	63–78 (G2)	85–110 (C5)	55–110
Alto	75–81 (C3)	108–115 (A5)	68–115
Dramatic soprano	55–88 (A3)	101–113 (D6)	55–120
Coloratura soprano	54–80 (B3)	110–118 (F6)	54–118

Directionality

As with the piano, the human voice’s ‘shadowing lid’ is represented by the head, the mouth position, and the funnel effect.^23,24 They strongly affect the sound propagation of the singer and prioritise the spherical area in front of the singer from −30° to 60° in a vertical plane and from 90° to 90° in a horizontal one. These angle ranges are referred to frequencies above 4000 Hz because the lower the frequency and more spherical the diffusion of the sound from the singer, the pressure level drops only behind the singer (i.e. at 250 Hz the sound diffusion is almost a complete sphere but the intensity drops 10 dB behind the singer).

This analysis led to the following:

Schubert’s friend and singer, Johann Michael Vogl, was a baritone, so the previous section could focus on his possible sound intensity extension according to his voice type. However, it is also useful to know the singer’s sound emission extension in regards to the intimate character of the art song, especially in the lowest pressure level.

Figure 2 shows how the singing human voice is very directional and depends on the singer’s posture orientation. This suggests that questions regarding the central position of the singer in the hall during the original performances must be considered.

Figure 2.

Directional characteristics of the singing voice for different vowels and angular regions of the singing voice in octave bands.⁵⁴

A socio-historical point of view

It may be useful to consider different methods of finding the original acoustical parameters’ values alongside significant social and cultural events that occurred in the first part of the 19th century in the music world. With the death of Beethoven in 1827 and Schubert’s death shortly thereafter,²⁵ Vienna lost its status as a prominent European musical capital whose famous musicians and figures founded many institutions in France, Germany, and England. Thus, we can observe how Viennese musical heritages began to spread across the rest of Europe at this time. For example, England soon became an important music centre, as many successful musicians were there. London in particular became a popular site for touring musicians of the time who were playing music with Viennese features. For that reason, it may be important to analyse the English concert halls as an extension of the musical tastes and acoustics of the Viennese period, as well as that of Schubert’s lifetime. To highlight this aspect, the foundation of a ‘Beethoven Quartet Society’ in late 1830 by a group of amateurs to facilitate ad hoc concerts of chamber music was remarkable.²⁶ Berlioz himself talked about them after an England trip. This is further proof of a possible spread of all the Viennese acoustical features in the new musical concert culture that the British industrial revolution brought about in the 19th century. Unfortunately, however, there are not sufficient data to suggest that the typical reverberation time of the Austrian theatres and music rooms was repurposed inside the British music halls, but it can surmised that Vienna’s music world did indeed influence many aspects of the music.

Acoustical parameters

The main goal of this paper is to understand what the original acoustic parameters for Lieder were in the early 19th century and to verify if the resulting values currently exist in contemporary halls for chamber music. In the previous paragraphs, the characteristics of the genre and the protagonists (piano and voice) were explained to suggest that the possible final result might be different when compared to our usual point of view. It was realised that the Lieder praxis indicates a general softer sound and a complex directional sound diffusion of both the piano and singing voice.

Of course, the criteria summarised in ISO 3382-1²⁷ can represent a complete way to analyse an art song performance. These parameters extracted from²⁷ are the results of research that was conducted to understand the behaviour of the sound inside existing, typical concert halls²⁸ in order to outline useful tools for managing the sound’s energy and to archive a complete control for the design of the listening experience.^11,29,30 The following equations concern how the measured parameters were calculated using RAMSETE software simulation in the present paper.³¹

Reverberation time (RT) is probably the most common parameter used to probe sound behaviour within a room. It represents the duration required for the space-averaged sound energy density in an enclosure to decrease by 60 dB after the source emission has stopped.

Because we are looking for values concerning music listening, reverberation time shows how an enclosed place influences a performance of a musical composition only at the end of the piece or in a particular set of moments characterised by long breaks when the sound stops.³² Because this is not sufficient for our analysis, we also consider Early Decay Time (EDT), the time it takes for the impulse response to initially decrease by 10 dB. However, such an event is more prone to happen in a music listening session, such as when the relation of each note to the subsequent note is measured.

The Clarity index $C_{80}$ or Early-to-Late coefficient, as seen in equation,¹ has an important role in the analysis of a mixed musical and text listening. It represents the ratio between the early sound’s reflection (in the first 80 ms) and the late reflections (from 80 ms on). Differences in the Clarity index’s values were usually considered for a better musical listening against a better speech understanding.³³ For this reason, $C_{80}$ is critical in our analysis.

C_{80} = 10 \log (\frac{\int_{0}^{80} p^{2} (t) d t}{\int_{80}^{\infty} p^{2} (t) d t}) [d B]

(1)

It is important to highlight that $C_{80}$ is evaluated only in the octave range between 125 and 4000 Hz. Despite the low influence that low frequencies have on Clarity,³⁴ the low range between 125 to 500 Hz is important in a Lied because it is a frequency range musicians are interested in during a performance. Furthermore, the dodecahedral source acquires an appreciable directionality above 4000 Hz, which makes it impossible to confute the theoretically calculated data with in situ recorded data³⁵; for this reason, 4000 Hz is the higher border of the analysed frequencies. 4000 Hz is also the highest range for available absorption coefficients data for some of the materials involved, as well as the highest frequencies of the sources.

The importance of the song’s text and its relation to the musical score was discussed earlier in Section 2.1. For this reason, the ‘Deutlichkeit’ (definition), D, the ratio between the integrated squared sound impulse in the first 50 ms and the integrated squared sound impulse during the entire time, was also analysed

D_{50} = \frac{\int_{0}^{50} p^{2} (t) d t}{\int_{0}^{\infty} p^{2} (t) d t} [%]

(2)

This can be useful to prove if a high level of D does not interfere with a musical performance that has a high Clarity index and a possibly ‘dry’ acoustic. This could lead to a conclusion about the importance of the articulation of the sung text, regarding whether it concerns only the performers, who have to sing the words as clearly as possible, or if the original conditions help the performers with this difficult task instead. For the same reason, it could also be useful to make a comparison between the Clarity index $C_{80}$ and the same parameter integrated in a different time interval of 50 ms. This parameter, $C_{50}$ , could highlight the resulting definition and the speech conditions of the hall.³⁶

Since our analysis helps to improve tools for hall design, how the architecture of the room influences the sound’s intensity should also be considered. In fact, architectural geometry can make the same acoustical source louder or softer. For this reason, we consider the Sound Strength G as the intensity level of the ratio between the squared impulse response for the entire time and the energy reaching a listener 10 m away under anechoic conditions.²⁷

G = 10 \log (\frac{\int_{0}^{\infty} p^{2} (t) d t}{\int_{0}^{\infty} p_{10}^{2} (t) d t}) [d B]

(3)

Another important parameter besides EDT that can have a key role in our dissertation is the Initial Time Delay Gap (ITDG), which is also known as Intimacy, named so by Beranek, and involves the sounds’ first reflection and its correlated energy.³⁷ ITDG refers to the time elapsed between the direct sound and the first reflection. Intimacy is an apt name choice for the parameter, as it represents perfectly what it entails, and should be a powerful tool for analysing a music genre in which instrument and voice are based on pianissimo dynamic levels and traditionally performed in domestic (although very wide) rooms. It seems that the more the hall expresses this Intimacy, the better it is suited for Lieder. Although Beranek calculated this parameter at a point barycentric to the hall, it can be derived from each impulse response. He also summarised in his 2004 book that, ‘in the best halls the ITDG is less than 25 ms’.

Another point that could be significant in our analysis is the relationship between a performer’s seen movement and the sound heard by the listener. The closer they are each other, the closer the listener will feel to the performer, which is emphasized by how clearly and easily the listener can look at the performer. However, this hypothesis is not new; the awareness of this aspect, which we may call listener-to-performer feeling, exists in other forms.³⁸ While it may not have been his intention, the American critic Neher,³⁹ during his own attendance at the art song recital by DiDonatao, confirmed that it is important that the listener’s distance from the performers is close enough for the listener to see the verbal expressions of the singer. He found that when he changed his position in the audience and moved closer to the stage, his experience improved.

The experiment: Recreate the hall where Schubert performed

Because it is very difficult to find the possible original places mentioned in Section 2.2, a live instrumental analysis is hardly applicable; however, modern instruments and technologies can help research studies discover interesting results. In this case, it could be useful to rebuild one of the halls where Schubert supposedly used to play.

‘An Evening at Baron von Spaun’

As said previously, it is quite difficult to find some of the halls from the list in Section 2.2 because an actual address of the venue is impossible to locate. While we cannot look to photographs for actual images of any of the locations since photography was not yet invented, we can look to paintings.

Moritz Von Schwind (1804–1871) was a young painter who often made Schubert and his music the subject of his art. He began working with the composer in 1821, providing illustrations for many of his songs.^14,40,41 In particular, there is a painting, ‘An Evening at Baron Von Spaun’ (Figure 3(a)), that depicts a Schubertiade in the apartment of Baron Von Spaun at Linz. The painting was completed in 1868 and was drawn from memory. The particularity and richness of the art, together with the painter’s attention to historical details,⁴² allow us to suppose that the scene is a realistic portrayal of an original Schubertiade. Since the painting is drawn using an accidental perspective, it is then possible to track the two vanishing points (Figure 3(b)) and recreate the plan and elevations of the part of the room in view (Figure 4).

Figure 3.

(a) Moritz Von Schwind’s ‘An Evening at Baron Von Spaun’ or ‘Schubertiade 1868’ (1868) drawn from memory. The picture shows Franz Schubert on the piano, as well as Josef von Spaun, Johann Michael Vogl, Franz Lachner, Moritz von Schwind, Wilhelm August Rieder, Leopold Kupelwieser, and, Eduard von Bauernfeld and (b) is a reconstruction of the perspectival scheme of the painting.

Figure 4.

(a) is a plan and elevation of the considered area on the painting. The lighter area is the suggested complementary part of the drawn side of the room, (b) is a rendering of the reconstructed room (V-ray for Rhinoceros software), and (c) is a picture of Johann Fritz’s piano (Vienna, 1811) (https://www.chrismaene.be/nl/historical-keyboard-instruments/pianoforte-johann-fritz/).

Given this information, it is necessary to state the following hypotheses:

Von Schwind’s painting is considered a fairly realistic portrayal of the original Schubertiade, since he attended many of these events and had a close relationship with the composer.

The interior design of the hall is recognized as a Biedermeier style,⁴³ which was popular at the time (1822–1830).

The dimensions of the hall are rectangular, as were the common aristocracy-owned buildings of the time.⁴⁴ The major side dimensions were derived by the position of the imaginary painter, assuming he took the farthest position away from the scene as possible. Moreover, the Austrian Biedermeier curved stove in the far right of the painting was used to dictate the corner of a room in order to prove the right-side wall’s position⁴⁵ (Figure 5).

The result is a 105 $m^{2}$ room with a ceiling height of 3.4 m and a complete volume of 357 $m^{3} .$

Figure 5.

(a) and (b) indicate the ceramic stove position in the Moritz Von Schwind painting, (c) is a typical Austrian Biedermeier ‘Kachelofen’ (ceramic stove), and (d) is an example of an Austrian Biedermeier curved stove example.⁴⁵

The acoustic of the ideal hall

In order to make a computational simulation based on the geometry of the hall, it is necessary to first know the materials of the room.

As stated previously, Lieder performances occurred at Von Spaun’s apartment during the second decade of the 19th century, when the interior design trend was Biedermeier, which is easily recognisable by the chairs, the floor, and the simple decorative details on the wall and the wallpaper, as well as the furnishings and piano. The presence of four anterior legs suggests that it could be a Viennese piano built by Johann Fritz (Figure 4(c)). The wallpaper was a standard requirement for any home inhabited by respectable middle/high-class individuals at the time, and was even considered a symbol of wellness.⁴³ There are also many identifiable similarities between this room and the Green Salon at the Wittumspalais of Duchess Anna Amalia in Weimar.⁴⁴ Furthermore, thanks to the work of Sangl,⁴³ it is possible to set the materials of the room and even predict possible furniture locations in the part of the room that is missing from the painting.

Indeed, much of the interior design trends of the time were controlled by rigorous rules. For example, it is even possible to know the probable colour of the wallpaper by interpreting the most followed interior-design book of the period, Theory of Colour by Goethe in 1810, in which green wallpapers are recommended for the most attended rooms of the house.⁴⁶ It is also possible to suggest the presence of a carpet. A particular kind of carpet manufactured in Linz, the town where the drawn scene was remembered, became very popular in 1810, but it had to be installed in the centre of the hall; in the painting, however, there is no presence of it.

The typical Biedermeier floor is wooden with elaborate decorations, like the interiors at the Schloss Rosenau from the same time period. There had to be a chandelier, but no further attention will be paid to it in this study because its metal components and small dimensions should not influence sound absorption.

The roof is assumed to be covered wallpaper because that was very common at the time.

Hence, the acoustical absorption coefficient of each material is derived from data found in the literature^47
–50 and RAMSETE software database (Table 4). The analyses were performed in the 125–4000 Hz interval due to the lack of data at the lower and higher frequencies.

Table 4.

Sound absorption coefficients in octave band range 125–4000 Hz for each material concerned the simulation. References for the data are reported (data from RAMSETE software materials database are reported with (R)).

Material		Sound absorption coefficient α
Material		125 Hz	250 Hz	500 Hz	1000 Hz	2000 Hz	4000 Hz
Plaster on masonry wall with wall paper on backing paper	(50)	0.02	0.03	0.04	0.05	0.07	0.08
Parquet on counterfloor	(50)	0.20	0.15	0.10	0.10	0.05	0.10
Pine wood	(R)	0.10	0.10	0.10	0.09	0.10	0.12
Audience in moderately upholstered chairs 0.90 m × 0.55 m	(49)	0.55	0.86	0.83	0.87	0.90	0.87
Unoccupied, moderately upholstered chairs 0.90 m × 0.55 m	(49)	0.44	0.56	0.67	0.74	0.83	0.87
Oil canvas	(47)	0.01	0.01	0.01	0.10	0.20	0.45
6 mm glass	(50)	0.10	0.06	0.04	0.03	0.02	0.02
Wood hollowcore door	(49)	0.30	0.25	0.15	0.10	0.10	0.07
Cotton curtains (0.5 kg/m²)	(49)	0.30	0.45	0.65	0.56	0.59	0.71
Glass window 4 mm	(50)	0.30	0.20	0.10	0.07	0.05	0.02
Ceramic tiles (Stove)	(50)	0.01	0.01	0.01	0.02	0.02	0.02

Figure 6 demonstrates that only the sited audience and curtains have significant sound absorption behaviour; all of the other surfaces have a low absorption coefficient (< 0.3) except for the paintings when the sound is over 2000 Hz. However, when the sound absorption coefficients are converted into sound absorption units by multiplying each coefficient value for the surface area, the following data are deduced (Table 5). In Figure 7, the greatest contributions are given by the parquet on the floor for the low frequencies and by the wallpaper for the high frequencies. The material distribution is also graphically represented in the related figures to Figure 8. The presence of the audience (23 chairs in the simulation case) has an important influence but it is not the most influencing parameter outside the 500–100 Hz frequency band.

Figure 6.

Sound absorption coefficients graphical representation in octave band range 125–4000 Hz for each material in the simulation. References for the data are reported (data from RAMSETE software materials database are reported with [R]).

Figure 7.

Graphical representation of sound absorption units in octave band range 125–4000 Hz for each material. References for the data are reported (data from RAMSETE software materials database are reported with [R]).

Figure 8.

Graphical distribution of the material inside the 3D room used for simulations. Colours’ materials are the same as in Figure 7.

Table 5.

Sound absorption units, surface and percentage on the total surface of each material in an octave band range of 125–4000 Hz. References of the data are reported (data from RAMSETE software materials database are reported with [R]).

Materials		Sound absorption unit						Surface [m²]	% on total surface
Materials		125 Hz	250 Hz	500 Hz	1000 Hz	2000 Hz	4000 Hz	tot. surface = 417.02 m²
Plaster on masonry wall with wall paper on backing paper	(50)	4.501	6.7515	9.002	11.2525	15.7535	18.004	225.05	53.97
Parquet on counterfloor	(50)	21	15.75	10.5	10.5	5.25	10.5	105	25.18
Pine wood	(R)	1.7	1.7	1.7	1.53	1.7	2.04	17	4.08
Audience in moderately upholstered chairs 0.90 m × 0.55 m	(49)	6.325	9.89	9.545	10.005	10.35	10.005	11.5	2.76
Unoccupied, moderately upholstered chairs 0.90 m × 0.55 m	(49)	5.06	6.44	7.705	8.51	9.545	10.005	11.5	2.76
Oil canvas	(47)	0.0608	0.0608	0.0608	0.608	1.216	2.736	6.08	1.46
6 mm glass	(50)	0.399	0.2394	0.1596	0.1197	0.0798	0.0798	3.99	0.96
Wood hollowcore door	(49)	2.46	2.05	1.23	0.82	0.82	0.574	8.2	1.97
Cotton curtains (0.5 kg/m²)	(49)	2.823	4.2345	6.1165	5.2696	5.5519	6.6811	9.41	2.26
Glass window 4 mm	(50)	5.385	3.59	1.795	1.2565	0.8975	0.359	17.95	4.30
Ceramic tiles (Stove)	(50)	0.08	0.08	0.08	0.16	0.16	0.16	8	1.92

Simulations and results

Computational simulations were carried out using RAMSETE software. The software has the ability to use a pyramid ray tracing method rather than a conic trade method in order to diffuse signal from a virtual source. This helps to avoid overlap in the signal-diffusion area and optimises the calculation processes.³¹ From a 3D cad, it is possible to assign different materials (with their own absorption features previously inserted into the material manager section) for each internal and exposed surface of the room. Then, once the source and receiver are located, the simulation can begin through the tracer section. Once this is complete, it is possible to view, gather, and acquire the output data into the Ramsete view section. Receivers are virtual omnidirectional microphones, and sources can be chosen or even created in the source manager section. The possibility to switch sources allows two different kinds of simulations: the first used an omnidirectional 100 dB white-noise source (Figure 5(a) and (b)), while the second used the coupled presence of a grand piano and singer source (Figure 5(c) and (d)), both extracted by the software source database. It should be noted that the second analysis contains a deliberate mistake: in the reconstructed hall of the painting, there is a Viennese action piano, which has quite different acoustic characteristics from a grand piano (Section 2.3). However, because a complete acoustic source analysis of an original Viennese action piano does not exist, the grand piano source is the closest representation that is currently available. Furthermore, although we are focusing on a historical acoustic recreation, a grand piano is the most common piano currently housed in any kind of hall, so analysing the behaviour of this younger but far more widespread instrument is in our best interest. There is also a third kind of simulation that considers the performer’s source but without any public attendance. This is used in specific cases to analyse the influence of the public’s presence on the room’s acoustic and is then compared to the previous simulation set, which includes the public. The analyses are focused on the 125–4000 Hz range due to the sound absorption data and also because it corresponds to the frequency range of the grand piano and sources of the singer. All the following analysed parameters refer to Section 3. The virtual omnidirectional source was located over the piano location at a height of 1.2 m. The grand piano source was located over the instrument at a height of 1 m and the singer source was located in relation to the painting, at the right of the piano and directly through receivers 1–2. The choice of direction is arbitrary, so one of the possible directions the singer can look towards during the performance was chosen (Figure 4(c)). The directivity parameters of the sources refer to the previous Section 2.4 and the source manager data of the computational software. The receivers were organised in a regular 3 × 7 grid and set 1.2 m high in order to cover all of the room. Although placing a receiver too close to the sources to avoid sources’ directionality influences is not recommended, this is not a concert hall and the dimensions are very different; therefore, a source’s directionality could influence the major part of the room. For this reason, the problem of placing a receiver close to the sources was disregarded. When reported, an average of calculated data from the receiver 10–12–13–14–15–16–17–18 are considered public data because the receivers covered the entire Main Listening Area (MLA) (Figure 9(a) to (c)). Due to the small dimensions of the room, a discussion of the study will highlight only the parameters with a just notable difference (JND) in order to avoid getting lost in minimal details. The JND values of each analysed parameter, which come from existing literature,^27,51 are reported in Table 6.

Figure 9.

RAMSETE software set-up: (a) is the plan view of the omnidirectional source ( A) configuration, which is coloured green, (b) is the axonometric view of the omnidirectional source (A) configuration, (c) is the plan view of the grand piano (A) and singer (B) source configuration, both in blue. The closed-lid piano and the singer direction (arrow) are represented by their shapes (dashed shapes because they do not exist in the model used in computational simulations), and (d) is the axonometric view of the grand piano (A) and singer (B) source configuration. The receiver points are numbered from 1 to 21. The Main Listening Area (MLA) is highlighted in red.

Table 6.

Just Notable Difference’s (JND) values used.

	T60	EDT	C80	D50	G
JND	5%	5%	1 dB	5%	1 dB

Reverberation time

Table 7 shows the average results in all the three simulations, which are all expected for a small room, although there are few surfaces with a high absorption coefficient. The reverberation time is short, with an apex of 1.25 s at low frequencies in the case of the performance-sources-configuration. Figure 10 graphically shows the small difference between an omnidirectional source and the presence of the performers’ sources. These differences can be disregarded because they comprise less than 4.2% of the reverberation value and are therefore not audible, even at the 500–1000 Hz range in both the case that a public audience is present or absent.

Table 7.

Average reverberation time (RT) of the room in the 125–4000 Hz octave band range in the three simulation configurations.

Configuration	Frequency (Hz)
Configuration	125	250	500	1000	2000	4000
Omnidirectional source	1.24	1.16	1.22	1.19	1.14	0.92
Performance sources	1.25	1.19	1.23	1.2	1.21	0.96
Performance sources (whithout public)	1.2	1.11	1.12	1.1	1.15	0.92

Figure 10.

Average room’s reverberation times (RT) graphical representation in the 125–4000 Hz octave band range in the three simulation configurations.

Early decay time (EDT)

As said previously in section 3, an analysis of Early Decay Time could provide more information about reverberation during the listening interval more so than during the reverberation time, which only becomes useful at the end of the piece or after significant pauses. Comparing the EDT charts (500 Hz) in Figure 11 shows the different behaviour of the reverberance depending on the source. In the omnidirectional case, the maximum value of the EDT reaches 0.9 s only in the area of receiver 11. An asymmetry arises from the analysis and a longer EDT is calculated on the left side of the room as compared to the right one. The difference is not notable (only 0.02 s), however, and probably caused by the denser presence of a public audience on the right side versus the left side. This supposition is confirmed by comparing Figure 11(b) and (c); although natural sources are more directive than omnidirectional ones, the absence of the public rebalances the EDT length in the two sides of the room. The public’s presence and position clearly influences the parameter in the hall but does not affect it: the differences are all less than the 5%. We can consider an original EDT value of 0.95 for performers (receiver 11) and a value of 0.88 for the public in the MLA.³²

Figure 11.

Early decay time maps of the room for omnidirectional source configuration (a), for grand piano and singer configuration (b), and for grand piano and singer configuration without public (c). Higher values are in red and lower values are in blue, and the frequency analysed was 500 Hz.

Clarity index $C_{80}$

Table 8 shows the Clarity indexes of the room. Since there is a great difference between the side where the singer is looking towards and the opposite position, a great difference between data is expected, as calculated by a virtual receiver close to the singer source and located at the point where the singer is facing. This difference is deduced by comparing the omnidirectional source and performances source charts in Figure 12: looking at receivers 11 and 14, their comparable index level in Figure 12(a) becomes 4.09 dB different than Figure 12(b). This is because the directionality of the singing voice, combined with the additional affection of the piano source. What is unexpected is the index level at receiver 12 location in the performance source’s case. This anomaly is not caused by the presence of the public, however, because the same phenomenon appears in Figure 12(c). Between receiver 12 and its symmetrical one, 14, there is a difference of 3.42 dB. By changing the integration time, the shorter 50 ms time of the $C_{50}$ parameter (Figure 13), a significant difference of 3.72 dB still exists, which is even greater than $C_{80}$ . This probably suggests that the smaller time interval that is considered (and closer to the direct sound we move), the bigger the energy ratio becomes. The cause of this asymmetry may be the instrumental sources; for this reason, the impulse response at position 12 is considered (Figure 14). The impulse response indicates how the major sound pressure level comes from the first interaction between sound wave and receiver rather than from the sound wave reflected from the environment. This consideration suggests that the instrumental sources are the cause of the anomaly.

Table 8.

Clarity index of the room in the 125–4000 Hz octave band range in the three simulation configurations. The performance source configuration (grand piano + singer) was considered in two opposite locations (receivers 1 and 4) according on the piano’s location due to the large difference that performance sources cause on the parameter.

Configuration	Frequency (Hz)
Configuration	125	250	500	1000	2000	4000
Omnidirectional source	5.59	6.06	5.83	5.99	5.7	7.03
Performance sources	5.47	5.91	5.71	5.82	5.7	6.87
Performance sources (whithout public)	4.9	5.19	5.36	5.56	5.63	6.98

Figure 12.

Clarity index $C_{80}$ maps of the room for omnidirectional source configuration (a), for grand piano and singer configuration (b), and for grand piano and singer configuration without public (c). Higher values are in red and lower values are in blue and the frequency analysed was 500 Hz.

Figure 13.

Clarity index $C_{50}$ maps of the room for omnidirectional source configuration (a), for grand piano and singer configuration (b), and for grand piano and singer configuration without public (c). Higher values are in red and lower values are in blue and the frequency analysed was 500 Hz.

Figure 14.

The performance (grand piano + singer) configuration’s impulse response at receiver 12.

It is also important to be aware of the high values of the early-to-late indexes of these simulations. In fact, looking at Table 8, the clarity index values were very high in all cases (omnidirectional and performance sources), well over the recommended values for music listening [−4 dB < $C_{80}$ < 2 dB³³] or for concert halls [−2 dB < $C_{80}$ < 2 dB] but, looking at the $C_{50}$ chart, values represent common situation in rooms for speech [inside −3 dB < $C_{50}$ < 9 dB range⁵²] We can thus assume an original Clarity index value of 10.5 dB for the performers and a value of 5.67 dB for the audience in MLA.

‘Deutlichkeit’ $D_{50}$

Table 9 indicates the $D_{50}$ parameter’s average values calculated inside the room. Although the definition index is somehow linked to the $C_{50}$ index by the integration time, $C_{50}$ is different from $D_{50}$ because it is a logarithmic ratio of early-to-late arriving sound during the first 50 ms, the other parameter is a linear ratio instead (Section 3). Anyway, no new important characteristics are introduced. Furthermore, no notable differences exist in the three simulation cases in the 125–4000 Hz range (Table 9). What is interesting to report, however, is the average value of the indexes: in each of the three analysed cases, the index’s values are beyond 50% over the suggested upper limit of the index for music listening and in the first 20% (between 50 and 70%) of the index’s value for optimal speech listening.³³ ‘Deutlichkeit’ confirms what was observed in Section 5.3.

Table 9.

‘Deutlichkeit’ index of the room (average) in the 125–4000 Hz octave band range in the three simulation configurations.

Configuration	Frequency (Hz)
Configuration	125	250	500	1000	2000	4000
Omnidirectional source	65.49	67.47	66.34	66.99	65.72	71.16
Performance sources	63.1	66.44	65.84	66.5	66.13	70.28
Performance sources (whithout public)	60.61	63.17	64.17	65.39	65.67	70.52

Sound strength G

Through an analysis of the sound strength index inside the room at the different receivers’ location (Figure 15), the same anomaly is reported in clarity and definition indexes (Sections 5.3 and 5.4). The public’s presence or absence affects the results in a very small manner. The public’s absence enlarges the green and yellow values areas (Figure 14(c)), as compared to the public’s presence (Figure 15(b)). Considering a barycentric location of the MLA as representative, the strength at receiver 17 is 18.3 dB when occupied and 18.68 dB when unoccupied.

Figure 15.

Strength maps of the room for omnidirectional source configuration (a), for grand piano and singer configuration (b), and for grand piano and singer configuration without public (c). Higher values are in red and lower values are in blue, and the frequency analysed was 500 Hz.

The initial time delay gap

As defined in Section 3, ITDG is the time gap between the direct sound from the source and the first reflected wave. In order to analyse two opposite values and locations inside the main listening area, the sound signals from receivers 14 and 16 are reported during the first 50 ms (Figure 16).

Figure 16.

The performance (grand piano + singer) configuration’s signal at receivers 14 and 16 during the first 50 ms.

From the analyses of the received signals, it is possible to realise a maximum value of 17 ms (receiver 14) and a minimum value of 8 ms (receiver 16). The latter value is very small but unsurprising considering the small dimensions of the room and the smaller distance between listener and wall at receiver 16. Since receiver 14 is the barycentric point of MLA, 17 ms should also be taken as the most representative of the hall.

This whole set of analyses leads to the following statements:

The original reverberation time was probably located between 0.92–1.25 s in the 125–4000 Hz range (Table 7, performance sources).

Original values for EDT could be 0.95 s for performers and 0.88 s for the public (in MLA).

Clarity and definition strictly depend on the direction the performer is singing but they have values optimal for speech rather than for music listening: 10.5 dB of clarity index for performers, 5.67 dB for public (in MLA), and a definition between 63.1–70.28% in the 125–4000 Hz range (Table 9, performance sources).

EDT, clarity, and definition indexes are affected by the instrument and vocal sources’ disposition and direction.

The rectangular room’s geometry does not seem to improve the sound diffusion in the area around the piano and singer (i.e. the area where the audience is located).

The initial time delay gap in the hall has a maximum value of 17 ms.

All of the above statements were derived from the RAMSETE computational analysis made on a suggested room with justified dimensions and materials. We hope that this research has demonstrated how it could be possible to assume an original acoustic condition of original music performance by a multidisciplinary approach and how studying the original acoustic is more convincing than studies on the major musical venues of the time that still exist. As reported at the end of Section 2.2, we can access to a limited data set from in situ analysis. Moreover, referring to those venues could mislead our assumptions. While studying major theatres may be interesting, it does not take into account a specific condition, such as how musicians (Schubert included) traditionally performed chamber music in smaller venues. Indeed, theatres were too big and expensive to allow these kinds of performance; only great and remunerative works such as operas were performed there. As discussed in Section 2.1, however, musicians often adapted the music to the space in which it was performed. Recreating the acoustical conditions of the Schubertiade could thus allow a recreation of the same sound experience as the original performance.

Data comparison with existing chamber music halls

The previous experimental analysis indicated the possible acoustic characteristics of the place where the performances of Lieder used to take place. Although this is not the primary aim of this preliminary study, it is also worth comparing the original performance data with that of contemporary stages to discover if halls exist that would allow a listener to experience a performance identical to that of its original context, and what the differences are when this recreation is not a possibility. Thanks to the analysis and cataloguing work of Hidaka and Nishihara,⁵³ it is possible to consider the halls summarised in Table 10.

Table 10.

Chamber music concert halls data from.⁵³ The listing is alphabetical. The subscripts ‘L’, ‘M’ and ‘3’ indicate, that the octave band average is f 125 and 250 Hz, 500 and 1000 Hz, and 500, 1000, and 2000 Hz, respectively.

Hall	V (m³)	N	RT_M,occ (s)	EDT_M,unocc (s)	C_80,unocc (dB)	G_L,unocc (dB)	G_M,unocc (dB)	ITDG (ms)
Amsterdam, Kleinersaal in Concertgebouw	2190	478	1.25	1.49	1.5	13.7	12.9	17
Berlin, Kleinersaal in Konzerthaus	2150	440	1.08	1.32	2	12.2	10.9	12
Kanagawa, Higashitotsuka Hall	3576	482	1.2	1.21	1.9	5.4	8.7	18.5
Kirishima, Miyama Conceru	8475	770	1.84	1.8	−0.1	8.2	8.3	25.5
Prague, Martine Hall	2410	201	1.76	2.19	−1.9	12.7	12.6	15.5
Salzburg, Grossersaal in Mozarteum	4940	844	1.66	2.06	−1.6	9.9	9.6	19.5
Salzburg, Wiennersaal in Mozarteum	1070	209	1.11	1.33	1.7	14.9	14.3	15
Tokyo, Casals Hall	6060	511	1.67	1.79	−1.3	7.6	9.4	16.5
Tokyo, Dai-ichi Seimei Hall	6800	767	1.66	1.83	−0.1	9.8	10.8	24
Tokyo, Hamarikyu Asahi Hall	5800	552	1.67	1.82	0	7.1	8.7	14
Tokyo, Ishibashi memorial Hall	5450	662	1.7	1.84	−0.8	9.2	10.8	19.5
Tokyo, Mitaka Arts Center	5500	625	1.73	2.26	−2.2	9	11.1	16.5
Tokyo, Sumida Small Sized Hall	1460	252	0.93	1.08	2.8	8.1	10.6	8.5
Tokyo, Tsuda Hall	4500	490	1.33	1.42	0.8	7.6	10.7	17
Vienna, Brahmssaal	3390	604	1.63	2.37	−2.8	12.8	13.6	10
Vienna, Mozartsaal in Konzerthaus	3920	716	1.49	1.79	−0.2	11.6	10.8	20.5
Vienna, Schubertsaal in Konzerthaus	2800	336	1.98	2.54	−3.3	14.7	13.6	11.5
Zurich, Kleinersaal in Tonhall	3234	610	1.58	2.11	−1.8	14.1	13.2	13.5
Von Spaun apartments’ room	357	23	1.22	0.89	5.67 (MLA)	18.4	18.7	(8 to) 17

Unfortunately, it is possible to only compare the parameters that do not depend strictly on the halls’ dimensions (like ITDG) because the small dimensions of the Von Spaun apartment’s room are not comparable to a space designed for a concert, even for the small audience of a chamber music recital. It is possible, however, to first highlight the large difference in terms of early-to-late decay. As shown previously, Lieder’s original acoustics matched speech listening requirements and the public value was 5.67 dB due to the small distance between performers and listeners; this decibel level is inapplicable to a modern concert hall and is well over the highest value of the Sumida Small Sized Hall in Tokyo, which is 2.8 dB.

RT (Figure 17) and EDT (Figure 18) reveal that there are halls that exist with similar characteristics to the original’s probable characteristics. The dark blue bars of the RT chart individualise the more similar to the original Von Spaun apartment’s room (VS). However, the RT of the VS could change depending on the number of audience members, which is not fixed. For this reason, we consider a confidence interval of 0.06 s, which is determined by simulating the absence or additional presence of 10 listeners in the room. Considering the VS analysis considered less furniture than what could have actually been present, halls with an $R T < 1.22 s$ are probably the most similar to the original (Figure 17, red outlined columns).

Figure 17.

Reverberation time RT comparison between the chamber music halls from⁵³ (in blue) and the Von Spaun apartment’s room (in orange). The halls highlighted in dark blue have a more similar RT than the VS (confidence interval 0.06 s). The halls with a RT < 1.22 s are highlighted in red.

Figure 18.

Intimacy or ITDG comparison between the chamber music halls from⁵³ (in blue) and the Von Spaun apartment’s room (in orange). The halls with a similar ITDG to the VS one are highlighted in yellow.

In regards to the Early Decay Time, 0.89 s is the average value at the MLA, with the omnidirectional source in unoccupied conditions at 500 Hz. The Sumida Small Sized Hall (yellow bar in Figure 18) has the closest value of EDT to the original. This hall was not considered ideal for its RT because it was too low; while it is one of the smaller halls on the list (1460 $m^{3})$ , the actual smallest is the Wiennersaal in Mozarteum at 1070 $m^{3}$ . This is important because it helps to avoid the idea that simply designing a small hall is sufficient to recreate the original conditions of an art song performance. What can be deducted from this information, however, is that it was easier to distinguish a sound in the possible original place where Schubertiades where performed, and most likely aided in understanding the sung text as well.

Although it is not possible to make an ITDG comparison since the parameter is too strictly linked to the architectural features of the space, it is notable that there are halls in the previous list, such as the Kleinersaal in Concertgebouw in Amsterdam and the Tsuda Hall in Tokyo, where Intimacy is comparable or even identical to the receiver 14 value of SV. Unfortunately, all the other reported parameters are completely different than SV; therefore, it is impossible to assume that we could live the original experience in these halls, as if we were seated in front of the piano while Schubert was playing.

Conclusions and future studies

This paper attempted to summarise the principal features of the art-song musical genre from musical and historical points of view. An acoustical introduction to the instrument and performers was given first, highlighting the historical differences. As a result of historical, architectural, and interior design analyses, a computational analysis of a possible original performing place of Schubert’s Lieder was made, suggesting the acoustical features of an original performance. Although the furniture and audience members could be different in number, the analysis on the dimensions, furniture, and materials of the hall still permitted further investigation on the subject. Early to late decay and definition indexes matched that of speech listening rather than musical listening, and Intimacy had a maximum duration of 17 ms. Comparisons to some aspects of European and Japanese chamber music concert halls from a database show different and similar values, especially for RT and EDT. It is interesting to note that there are a few halls that recorded the same ITDG of SV; however, many questions remain unsolved and future studies are required for different aspects surrounding the topic. The first concern is the instrument: this analysis was made using a grand piano source and not a Viennese action piano. Section 2.4.1 demonstrated why the piano was played with a closed lid, but Section 2.3 explained the important differences between the two piano technologies. For this reason, it is necessary to acquire technical and source data of the Viennese piano technology to check the possible differences during simulations. Another important aspect is to compare the original acoustic features with a Viennese action piano to a modern concert hall’s acoustic features when playing a modern grand piano. As mention in Section 2.4.2, piano technology changed when halls became bigger and more attuned to the musical genres that were being performed. This parallel and cause-effect development may have important results regarding Lieder as well.

The last aspect concerns Lieder directly. This paper tried to recreate the original performances, keeping in mind that composers often adapt their works to the acoustics of the venue where they are typically performed. However, one is inclined to wonder whether the original situation is really the best for listening to Lieder. For this reason, it will be important to make analyses on the stages where acclaimed Lieder are performed today, including the Wigmore Hall in London, a hall that has fascinated great artists and composers such Britten or Rubinstein with its incredible acoustic features⁵⁴ and where the first complete cycle of Schubert’s Lieder was performed in the UK, the theatre in Schwarzenberg, where the Schubertiade Schwarzenberg Franz Schubert festival is organised, the American Zankel Hall in Carnegie Hall, and the Alice Tully Hall in Lincoln Centre. All of these stages have hosted Lieder concerts that have been reviewed by the critic Neher³⁹ There it is probably allowed to explore the theme of the listener-to-performer feeling. This aspect was introduced in Section 3 but it was not examined in the present study because more related to a contemporary case in modern concert hall, where public is turned towards the stage and singer faces people during almost the performance, rather than in small hall where singer faced the audience at will.

Footnotes

Declaration of conflicting interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

P. Domenighini

References

Tronchin

Knight

DJ.

Transmitting acoustic phenomena and aural illusions: examples from Athanasius Kircher’s

Phonosophia anacamptica. Build Acoust 2018; 25: 101–110.

Schubert

Tzekakis

EG.

The ancient Greek theater and its acoustical quality for contemporary performances. J Acoust Soc Am 1999; 105(2): 1043.

Beranek

LL.

Analysis of Sabine and Eyring equations and their application to concert hall audience and chair absorption. J Acoust Soc Am 2006; 120(3): 1399–1410.

Cirillo

Martellotta

Acoustics of Apulian-Romanesque churches: correlations between architectural and acoustic parameters. Build Acoust 2003; 10: 55–76.

Pompoli

Prodi

Farnetani

A note on the acoustics of orchestra rehearsal rooms. Build Acoust 2012; 19: 1–12.

Kramer

The Schubert lied: romantic form and romantic consciousness [1986/2016]. In: Leppert

Bernhart

(eds) Song acts. Leiden: Brill Rodopi, 2017, pp.52–85. DOI: 10.1163/9789004342132_003.

Schleiermacher

FDE

Wojcik

Haas

The hermeneutics: outline of the 1819 lectures. New Lit Hist 1978; 10(1): 1–16.

Horton

Kramer

. The Schubert lied: romantic form and romantic consciousness. In: Schubert. London: Routledge, 2019, pp. 311–348. DOI: 10.4324/9781315088396-22.

Kramer

Franz Schubert: sexuality, subjectivity, song. Ger Stud Rev 2000; 23: 595.

10.

Meyer

Raumakustik und orchesterklang in den konzertsaelen Joseph Hadyns. Acta Acust United Acust 1978; 41(3): 145–162.

11.

Meyer

Acoustics and the performance of music. New York: Springer, 2009.

12.

Cox

The sound book: the science of the sonic wonders of the world. New York: W.W. Norton, 2014.

13.

Clericetti

Charles Marie Widor. La Francia organistica tra Otto e Novecento. Varese: Zecchini, 2010.

14.

Gibbs

CH.

The life of Schubert. Cambridge: Cambridge University Press, 2000.

15.

Kokits Mag. Schubert memorial places. Franz Schubert (1798-1828) - biography 2, http://www.martijnhooning.nl/muziekgeschiedenis/schubert_gedenken_e.html#top (2018, accessed 25 July 2020).

16.

Harer

Musical venues in Vienna, seventeenth century to the present. Perform Pract Rev 1995; 8(1): 83–92. DOI: 10.5642/perfpr.199508.01.08.

17.

Bilson

Schubert’s piano music and the pianos of his time. Stud Musicol Acad Sci Hung 2017; 4: 263–271.

18.

Darmstädter

Huber

Hopfner

Das Wiener Klavier bis 1850: Bericht des Symposiums “Das Wiener Klavier bis 1850” veranstaltet von der Sammlung alter Musikinstrumente des Kunsthistorischen Museums Wien. Tutzing: Musikantiquariat und Verlag Hans Schneider, 2008.

19.

Bork

Akustische untersuchungen an klavieren und flüge. In: Ber. 17 Tonmeistertagung Karlsruhe 1992. Munchen: K.G. Saur, 1993, pp.751–764.

20.

Winckel

Auditory analysis of voice and speech sounds. Logos 1963; 6: 3–12.

21.

Sundberg

The acoustics of the singing voice. Sci Am 1977; 236(3): 82–84, 86, 88–91.

22.

Burghauser

Spelda

Die akustischen grundlagen der instrumentation. Regensburg, Germany: Gustav Bosse Verlag, 1971.

23.

Dunn

Farnsworth

DW.

Exploration of pressure field around the human head during speech. J Acoust Soc Am. 1939; 10(3): 184–199.

24.

Marshall

Meyer

Directivity and auditory impressions of singers. Acta Acust United Acust 1985; 58: 130–140.

25.

Biba

Schubert’s position in Viennese musical life. 19th Century Music 2019; 3(2): 106–113.

26.

Weber

Music and the middle class: the social structure of concert life in London, Paris and Vienna between 1830 and 1848. London: Routledge, 2003.

27.

BS EN ISO 3382-1:2009. Acoustics: Measurement of room acoustic parameters – part 1: Performance spaces (ISO 3382-1:2009). British Standard, Standards Policy and Strategy Committee. Brussels: European Committee for Standardization, 2009.

28.

Beranek

LL.

Concert halls and opera houses: music, acoustics, and architecture. 2nd ed. New York: Springer, 2005.

29.

Ando

Concert hall acoustics. Berlin: Springer, 1985.

30.

Ando

Raichel

DR.

Architectural acoustics: blending sound sources, sound fields, and listeners. J Acoust Soc Am. Epub ahead of print 1998. DOI: 10.1121/1.423953.

31.

Farina

. Pyramid tracing vs. ray tracing for the simulation of sound propagation in large rooms. In: Proceeding of the international conference on computational acoustics and its applications, COMACO95, Southampton, England, 1995.

32.

Barron

Interpretation of early decay times in concert auditoria. Acta Acust United Acust 1995; 81: 320–331.

33.

Bianchi

Carratù

L’acustica in architettura. Torino, Italy: CittàStudi, 2007.

34.

Beranek

Schultz

TJ.

Some recent experiences in the design and testing of concert halls with suspended panel arrays. Acustica 1965; 15: 307–316.

35.

Barren

Using the standard on objective measures for concert auditoria, ISO 3382, to give reliable results. Acoust Sci Technol 2005; 26: 162–169.

36.

Bradley

Reich

Norcross

SG.

A just noticeable difference in C50 for speech. Appl Acoust 1999; 58: 99–108.

37.

Hyde

JR.

Discussion of the relation between initial time delay gap (ITDG) and acoustical intimacy: Leo Beranek’s final thoughts on the subject, documented. Acoustics 2018; 1: 561–569.

38.

Pfordresher

PQ.

Coordination of perception and action in music performance. Adv Cogn Psychol. 2006; 2(2–3): 183–198.

39.

Neher

The art song recital in review. Hudson Rev 2011; 64(2): 325–330.

40.

Bisanz

RM.

Schwind, Moritz (Ludwig) von. In: Oxford art online. 2018. DOI: 10.1093/gao/9781884446054.article.T076948.

41.

Stahl

. Art books: A basic bibliography of monographs on artists by Wolfgang M. Freitag. Art Doc J Art Libr Soc North Am. Epub ahead of print 1986. DOI: 10.1086/adx.5.2.27947592.

42.

Seidel

Die Zauberflöte im bild der symphonie: Über ein aquarell von Moritz von Schwind. Munich, Germany: Bruckmann’s Verlag, 1882.

43.

Sigrid Sangl

. Biedermeier to Bauhaus. New York: Harry N. Abrams, 2001.

44.

Martin

Architectural space in eighteenth-century Europe:constructing identities and interiors. London: Routledge, 2017.

45.

Franz

Der Kachelofen: Entstehung Und Kunstgeschichtliche Entwicklung Vom Mittelalter Bis Zum Ausgang Des Klassizismus. Graz: Akademische Druck- und Verlagsanstalt, 1969.

46.

Wilson

Matthaei

Aach

, et al. Goethe’s Colour Theory. New York: Van Nostrand, Leonardo, 1972.

47.

Martellotta

Castiglione

. On the use of paintings and tapestries as sound absorbing materials. In: Proceedings of forum acusticum, Aalborg, Denmark, 27 June–1 July 2011, pp.1847–1850.

48.

Kelly

RM.

Historic paperhanging techniques: A bibliographic essay. SSRN Electronic Journal. 2012. DOI: 10.2139/ssrn.1986522.

49.

Vorländer

Auralization: fundamentals of acoustics, modelling, simulation, algorithms and acoustic virtual reality, vol. 53. Dordrecht: Springer, 2007, pp.1689–1699.

50.

Akustic. Absorption coefficients for generic building constructions, http://www.acoustic.ua/st/web_absorption_data_eng.pdf (2010, accessed 25 June 2020).

51.

Martellotta

Cirillo

Carbonari

, et al. Guidelines for acoustical measurements in churches. Appl Acoust 2009; 70(2): 378–388.

52.

Barron

Objective assessment of concert hall acoustics using Temporal Energy Analysis. Appl Acoust 2013; 74: 936–944.

53.

Hidaka

Nishihara

Objective evaluation of chamber-music halls in Europe and Japan. J Acoust Soc Am. 2004; 116(1): 357–372.

54.

Wulfrankz

Orlowski

RJ.

Acoustic analysis of Wigmore Hall, London, in the context of the 2004 refurbishment. Proc Inst Acoust 2006; 28: 255–267.

On best acoustical parameters’ values for ‘ Liederistic ’ music performance: A preliminary study

Abstract

Keywords

Introduction

State of the art

Schubert’s Lieder

A historical approach to the acoustics of Schubert’s time

The piano technology in Schubert’s time

The piano and the singing voice

The piano

Loudness

Directionality

The singing voice

Sound power

Directionality

A socio-historical point of view

Acoustical parameters

The experiment: Recreate the hall where Schubert performed

‘An Evening at Baron von Spaun’

The acoustic of the ideal hall

Simulations and results

Reverberation time

Early decay time (EDT)

Clarity index C 80

‘Deutlichkeit’ D 50

Sound strength G

The initial time delay gap

Data comparison with existing chamber music halls

Conclusions and future studies

Footnotes

Declaration of conflicting interests

Funding

ORCID iD

References

Clarity index $C_{80}$

‘Deutlichkeit’ $D_{50}$