Proceedings of the Institute of Acoustics

 

A comparison of musical performances in different acoustic environments created with an active acoustics system

 

Benedikt Roß, Amadeus Active Acoustics, Vienna, Austria
Volker Werner, Amadeus Active Acoustics, Vienna, Austria
Clemens Frischmann, Amadeus Active Acoustics, Vienna, Austria
Simon Neeten, Amadeus Active Acoustics, Vienna, Austria

 

1 INTRODUCTION

 

Musical performances are closely linked to the acoustics of the room in which they take place. The assumption, that playing music in a room with good acoustics is more enjoyable and that the musicians feel more comfortable, is obvious. But do the room acoustics also change the way the performers play? Do they influence the chosen tempo, the loudness and the range of musical dynamics? Many musicians claim to react consciously to the sound of the room in their interpretation. For example, conductors describe this circumstance in interviews with Beranek¹. But apart from these conscious adjustments, are there perhaps also subconscious changes that even the musicians themselves are unaware of?

 

To examine these questions in more detail, performances by a string quartet and a solo cellist were recorded and analyzed in a room with an Active Acoustic System. This setup allowed the performers to play the same piece in five different acoustic environments within just one hour. The evaluation assessed several parameters that can vary in musical performances, such as tempo, loudness and dynamic range. In addition, the musicians were asked about their own perception and these answers were also evaluated.

 

2 PREVIOUS RESEARCH

 

In recent decades connections between room acoustics and musical interpretation have been investigated. A study examining performances by several pianists in a room with variable acoustics demonstrated that longer reverberation times lead to a reduction in loudness². However, the reduction of the playing tempo with longer reverberation times, which is often described by musicians and seems to make sense depending on the character of the piece³, could not be proven. Only later studies were able to scientifically prove this phenomenon, whereby not only very long reverberation times seem to reduce the playing tempo, but also the dry acoustics of the anechoic chamber in which the recordings were made^4,5.

 

With their studies of concert performances by a solo cellist in various concert halls, Schärer Kalkandjiev and Weinzierl were able to demonstrate the influence of room acoustics on additional musical parameters, such as overall loudness, dynamic range and timbre⁶. The detailed analyses of the concert recordings revealed that different acoustic parameters influence different interpretative aspects more strongly than others. Also, results can be observed which vary from previous studies, for example the relationship between loudness and reverberation time. It is suspected that the reason for these contradictory results lies in the different individual preferences and levels of experience of the performers.

 

A later publication by Schärer Kalkandjiev and Weinzierl explained at least partially how these deviations arise: The experiment with different musicians in virtual acoustics could demonstrate that the interpreters' subjective assessment of the acoustic room quality can differ. Also, the perception of quality had an influence on certain aspects of the interpretation. Even though the musicians did not agree on, which rooms they found best, the result was clear: In rooms in which musicians feel more comfortable, they play faster, force less loudness, make more use of the dynamic range and use less agogic⁷. The finding that different musicians do not adapt their interpretation in the same way to changing room acoustic environments is also confirmed in other studies^5,8.

 

However, with all these experiments, there are always problems that arise from the fact that the acoustics of a room cannot be changed easily and quickly by conventional means. The research results obtained from the project with the cellist came from various concerts that took place on different days, which means that other factors besides the room acoustics, such as the mood and state of mind of the artist, could also have influenced the interpretation. In order to avoid this problem, later research projects worked with binaural impulse responses, which enabled the musicians to play in different virtual acoustics with headphones and without changing locations^7,9,10. Although the choice of the headphone system enabled the musicians to perceive the direct sound of their instruments without alteration, the unfamiliar situation could have influenced the interpretation. Moreover, the credibility of the perception of binaural audio simulations is strongly dependent on how well the simulated HRTFs correspond to those of the performer´s ears.

 

Therefore, the approach of the project presented and evaluated in this paper is a third one: the musicians were able to play in a room equipped with a temporary installation of an Amadeus Active Acoustics system. This made it possible to change the acoustics instantly at the push of a button without the musicians having to move to another room. At the same time, the performers could play without headphones and perform in a familiar setting. This offered an optimal starting point for such a sensitive and differentiated topic as the interpretation of music.

 

3 LOCATION

 

The recordings of the performances were all made as part of a thesis project in the “Klangtheater” Hall (“soundtheater”) at the University of Music and Performing Arts Vienna (mdw). The “Klangtheater”, in the further course referred to as “Dry Hall”, has a footprint of approximately 180 square meters and a height of 8.7 meters. Since it is primarily intended for the presentation of immersive electro-acoustic compositions and pop-music performances, the natural acoustics are kept dry. The average reverberation time is 0.5 seconds with an even increase of RT60 towards the low frequencies. This creates a slight imbalance, as the reverberation time in the bass frequencies is about twice as long as in the high frequencies (Figure 3). Based on this natural room acoustics four presets were created using Amadeus Active Acoustics to generate acoustics of rooms with different characteristics. To evaluate these acoustics, measurements were made according to ISO 3382.

 

 

Figure 1 and 2: The Klangtheater at mdw, Vienna. Implemented speaker setup differs from rendering.

 

In the first preset "Reflective" the acoustic dimensions of the room were not increased (the Amadeus Core calculates the natural time delays of the early reflections based on a geometric three dimensional model of the underlying natural room). The first reflections were enriched and amplified. The averaged RT60 of this preset was about 0.8 seconds, only slightly prolonged compared to the “Dry Hall”. The time-pattern of the early reflections of the second preset "Mozart" was based on a room size like the Mozart Hall at the “Konzerthaus” Vienna. Its acoustic properties were primarily designed for the performance of chamber music. The average RT60 of this preset was about 1.7 seconds. The third preset "Golden" was based on a room the size of the Golden Hall at the “Musikverein” in Vienna. The average reverberation time of this preset was about 2.2 seconds. Compared to the other presets, a slightly brighter color in the upper mid frequencies could be noted. The acoustics of the fourth preset “Cathedral” were oriented towards a large sacred space. The average RT60 in this case was 6.0 seconds. This setting was designed with a dark color in the decay. In all presets an increased reverberation time was implemented at low frequencies.

 

 

Figure 3 and 4: Reverberation times of the different acoustic environments: Figure 3 without the “Cathedral” preset for standard scaling, Figure 4 with all presets for comparison.

 

The EDT of the “Dry Hall”, the "Reflective" and the “Cathedral” preset was approximately in the range of the RT60. With the two presets "Mozart" and "Golden" it was clearly below the RT60 (Figure 5). However, due to the small size of the room, the distance between the sound source and some of the measurement microphones was rather small, which could at least partly explain the lower EDT.

 

The larger the simulated rooms were, the lower the clarity C80 became. This is a constant trend, from which only the "Golden" preset stood out: instead of dropping further, the clarity was almost the same as in the "Mozart" preset.

 

 

Figure 5: Averaged acoustic parameters for the different acoustics: RT60mid, EDTmid and TS (center time) are averaged for 500 Hz and 1 kHz. C50 and C80 are averaged for 500 Hz, 1 kHz and 2kHz. Bass Ratio indicates the RT60-ratio of 125 and 250 Hz versus 500 Hz and 1 kHz.

 

4 QUESTIONNAIRES

 

After their performance the musicians were asked to fill out a questionnaire. All the interviewees agreed that the acoustics had a strong influence on the way they felt when playing. Some of the musicians felt most comfortable in the "Reflective" acoustics and some in the "Mozart" acoustics. The "Cathedral" acoustics were described as the most uncomfortable by the members of the string quartet, on the grounds that they had to concentrate very much on their interaction. But the musicians also did not feel comfortable in the acoustics of the “Dry Hall” either.

 

However, the performers did not agree, in which acoustics it was easiest to play together. For all of them the “Golden” and “Cathedral” acoustics tended to cause more problems, as mutual hearing deteriorated, and more eye contact was necessary. The dry room, on the other hand, was partly perceived as pleasant for playing together, but partly also as difficult: After switching off the “Mozart” acoustics and then playing in the “Dry Hall”, the musicians immediately began to discuss whether they could hear each other better, whereby the “Mozart” acoustics tended to be more pleasant for those sitting further away and the “Dry Hall” for those sitting in the middle. A possible explanation for this could be the higher clarity in the “Dry Hall”, which facilitates interaction by making all the details of the other musicians clearly audible. On the other hand, the levels of the more distant instruments drop more quickly when there is less acoustic support, making them significantly quieter than the directly adjacent instruments. This makes playing more difficult for those performers sitting at the edge of the ensemble.

 

When asked if the acoustics had an influence on the interpretation, the solo cellist stated that he held a fermata in the Prelude longer in the "Cathedral" acoustics so that the reverberation of the room would not cover the beginning of the next section. He also felt that with longer reverberation times, a slower playing tempo might be appropriate and noted that the version he played in the very dry acoustics was his worst, he even had to stop and start again.

 

The responses of the members of the string quartet went in different directions. One violinist wrote that she had to press less in the more supportive acoustic environments and that it made sense to take more time for lengthened notes when the reverberation time is longer. The other violinist stated that she had to exaggerate her articulation in the "Cathedral" acoustics. The cellist of the quartet wrote: "Somehow I felt different about my dynamics. […] I played with less dynamic in E ["Cathedral"] and more dynamic as it goes towards A [“Dry Hall”].” During the recording, the quartet spoke only once about a conscious change in interpretation: after switching from the dry acoustics to "Reflective", the first violinist said that they could now “think in half bars” again.

 

5 INTERPRETATION COMPARISON

 

For the comparison of the interpretations, the instruments were picked up with supercardioid microphones at a distance of approx. 45 cm. Considering the directional characteristics of the microphone¹¹ the calculated level difference between direct and diffuse sound was between 23 dB ("Dry Hall") and 12 dB ("Cathedral"). The directional characteristics of the instruments that push the critical distance further away¹² could not be taken into account in these calculations. It is therefore assumed that the actual values are somewhat better than those calculated. Due to the small room volume and the long reverberation time of the “Cathedral” preset, it was not possible to achieve larger level differences without attaching the microphones to the instruments. This was avoided in order not to disturb the playing situation. Nevertheless, it is assumed that the decay of the room has no relevant influence on the recordings and thus on the evaluation of the loudness and dynamics comparisons. However, it cannot be ruled out that the values of the “Cathedral” preset are slightly influenced by it.

 

The solo cellist played the Prelude from Bach Suite Number 1 BWV 1007 in its entirety and the beginning of the Allemande. The string quartet played the beginning of the first movement from Mendelssohn's sixth string quartet, as well as the beginning of the first movement of Dvorak's eleventh string quartet. Before the recording, the performers were given the opportunity to listen to the different acoustics and rehearse in them. The pieces were recorded one after the other in the different acoustic environments with short pauses for regeneration and rehearsal in the new acoustics. The performers all played in the "Mozart" acoustics first and then successively in the other acoustics from dry to very reverberant (1. Mozart, 2. Dry Hall, 3. Reflective, 4. Golden, 5. Cathedral).

 

5.1 Tempo

 

For the evaluation of the playing tempo, the time in which the performers played the selected section was compared. The differentiated observation of smaller sections and conscious listening to the various recordings ensured that the results were not falsified by individual fermatas or pauses, but that there were actual changes in tempo. In the case of the Mendelssohn, a uniform development can be determined, which behaves in the same way as has been observed in other studies on this subject. In very short (“Dry Hall”) and very long reverberation times ("Cathedral" acoustics), the tempo slows down, while in the acoustics usual for concert halls, the playing tempo increases. But these differences are very slight and, if at all, on the edge of perceptibility. However, the tempo differences in some parts, such as the first eight bars, are more pronounced and therefore more likely to be noticeable, while behaving in the same way as the overall tempo-tendency.

 

With the Dvorak the playing tempo is very consistent and the same for all acoustics. The small deviations (less than 0.5%) are negligible. One can only speculate why no differences were found, but it cannot be ruled out that the piece or the tempo of the piece have an influence, as other studies have already suggested^5,8. Furthermore, it is assumed that it is more difficult for a string quartet than for a solo artist to intuitively adjust the playing tempo. In the above-mentioned conversation of the string quartet, the first violinist postulated during the Mendelssohn recordings, after changing the switched-off acoustics to the "Reflective" preset, that they could now think more in half bars again, which may have partly contributed to the slightly faster tempo.

 

The cellist's tempi do not reveal any clear tendencies. They are slightly inconsistent, without a systematic connection to the acoustics. However, the tempo chosen in the "Cathedral" acoustics is slower than in the other ones. This observation fits the cellist's statement and seems to be a conscious decision.

 

 

Figure 6: Deviations of the length of the performance from the average length in percent.

 

5.2 Loudness

 

To assess the loudness of the performances, the integrated LUFS values were calculated and compared. In addition, the same values were examined section by section to exclude the possibility of individual events distorting the overall result. In the Mendelssohn performances only slight deviations in loudness and no systematic correlation with the acoustical parameters can be observed. With the Dvorak the loudness increases slightly with increasing generated room size.

 

 

Figure 7: Deviations of the loudness (LUFS) from average loudness in dB.

 

A different trend can be observed in the cellist's performances. The overall loudness is highest at the "Mozart" preset and decreases towards the dry and livelier acoustics. Especially with the strong acoustical response of the "Golden" and "Cathedral" presets, the total loudness is one decibel lower than with the "Mozart" preset and thus perceptible. This could be because the performer already felt more supported by the room acoustics and therefore reduced his loudness. Or he wanted to trigger the room response less due to the lower clarity and longer reverberation time. As Schärer Kalkandjiev and Weinzierl have already suggested in one of their studies⁶ the different reactions could be caused by the different levels of experience of the musicians, if not simply due to individual taste. Relating to the quartet another possibility is that the increased volume is an attempt to counteract the more difficult conditions for hearing each other.

 

5.3 Range of musical dynamics

 

To assess the dynamic range of the interpretations, the parameter LRA (Loudness Range) was calculated, which is used in Mastering to determine the musical dynamics of a record. With this method, results could be shown regarding the dynamic range of the entire performances. The interpretations of both pieces by the string quartet reveal an increased loudness range in acoustic environments that exhibit characteristics of conventional concert halls. In the case of the Mendelssohn, the LRA value in the "Mozart" and "Golden" acoustics is almost a whole Loudness Unit above average. Remarkable is the strong outlier in the "Cathedral" acoustics, where the LRA is almost 2 LUs below average. For the Dvorak, the differences are less extreme, but are in line with the results of the analysis of the Mendelssohn recording. Even assuming that due to the small critical distance of the “Cathedral” acoustics the reverberation of the room has an influence on the recording and thus on the evaluation of the LRA values, it seems probable that these deviations do not produce such clear differences as can be found in the evaluation.

 

Also in the cellist's interpretations, the dynamic range varies in the different acoustics. The dynamic range decreases towards the “Dry Hall” and the “Cathedral” acoustics and is greatest with the "Mozart" preset. Only with the "Golden" preset the values are not as high as with the string quartet. Overall, the graphs show a clear tendency: The dynamic range increases in the acoustic environments with usual concert hall characteristics. In the more extreme acoustic conditions, both very lively and very dry, it decreases.

 

Figure 8: Deviations in LUs of the LRA from the average for the different acoustics.

 

This seems to be a subconscious phenomenon, as none of the musicians mentioned in the interviews that the acoustics influence the dynamic range in this way. Only the cellist of the quartet said that he had adjusted his dynamic behavior. However, he described an increase in dynamics towards the dry acoustics, which was not the case. Due to the correlating changes in the various acoustic parameters, it is not possible to say with certainty which one mainly influenced the reduction of dynamics. It seems reasonable to assume that in the very strong "Cathedral" acoustics, the musicians did not want to trigger the room as much and therefore played the loud passages a little softer. In addition, it is more difficult to play together tightly when the clarity is lower, which could explain the greatly reduced dynamics in the Mendelssohn. In the dry acoustics, again, the very low support for the performance could lead to somewhat louder piani, otherwise the sound would disappear completely.

 

5.4 Further Details

 

To check for further musical details and interpretive aspects that are difficult to detect with a computer analysis, the recordings were carefully compared aurally. This allowed further observations: It is obvious from careful listening that the acoustics influence the interplay. Already in the first eight bars of the Mendelssohn, it is noticeable that the quartet takes longer to find the same tempo in the "Cathedral" acoustics and even then, does not play together as precisely as it does in the “Dry Hall” and the "Reflective" acoustics. Also, in cases where one of the musicians gets a little out of rhythm during the performance, it usually takes less time in the drier acoustics until a clean interplay is restored. This phenomenon can most likely be explained by the lower clarity values.

 

In addition, at some spots the articulation in the violins might be a little bit more pronounced in the "Cathedral" and “Golden” acoustics, but not as much as one might expect based on the questionnaire. Further adaptations can be observed in a ritardando, in which the performers take more time with increasing reverberation time.

 

The most obvious difference in the solo cellist's performance is that a fermata at the end of one section of the Prelude was sustained much longer in the "Cathedral" acoustics before the performer began the next part. This also corresponds with the cellist's statements and is therefore certainly a conscious decision based on the reverberation time. In some places you can hear that in the Prelude the 16th notes were more accentuated and less connected in the "Cathedral" acoustics, which is an understandable reaction considering the reduced clarity. Finally, the tempo maxima of the rhythmically overall very free interpretation are somewhat reduced in the "Cathedral" preset compared to the other versions, which means that the acoustics also have an influence on the agogics.

 

6 CONCLUSION

 

Even if an adaptation to the changed acoustic environment could not be proven for all examined musical parameters, there are several areas in which a systematic change could be observed.

 

The most consistent results were found in the investigation of the musical dynamic range. Performers tend to expand the dynamic range in acoustics with typical concert hall properties, whereas the dynamic range is restricted in very dry or very lively halls. According to the questionnaires this seems to be a subconscious reaction. It can therefore be assumed that this trend is strongly dependent on the subjective feeling of comfort and security of the performers, so that more is risked in quiet piano and loud forte passages.

 

The influence of the acoustics on the playing tempo was shown in the interpretations of the string quartet by Mendelssohn in the way that has already been demonstrated in other research papers, albeit only to a very slight extent. The cellist also reacted to the reverberation time with his playing tempo, but only in the "Cathedral" acoustics. Consequently, it can be assumed that in musical interpretations, depending on the artist and the piece, a reduction in playing tempo can occur with longer and, in some cases, shorter reverberation times.

 

The effects of acoustics on loudness could be observed to a small degree. Here, the reaction of the quartet differs from that of the cellist. When examining the influence of acoustics on loudness, the reactions of performers are very individual, and no clear tendency can be discerned.

 

Looking at the musical details through analytical listening, other differences could be found. The reduced clarity can lead to prolonged fermatas, more accentuated articulation, and difficulties in interplay, which result in rhythmic inaccuracies. All these aspects coincide with the adaptations described by the musicians themselves, which they attribute to the acoustics.

 

To further differentiate the results of this research, it would be best to work with a bigger number of solo artists at different levels of training. It is hypothesized that increased experience will influence how musicians respond to a hall¹³. When working with ensembles, it is important to remember that spontaneous adjustments to interpretation are somewhat more difficult than with soloists, as much more communication is required. This is especially true for aspects of rhythmic interplay such as tempo and agogic. That's why it would be interesting to work with a very experienced ensemble that can react more intuitively and already has more experience in dealing with different room acoustics.

 

To be able to work with more instrumentalists, it would be advantageous to use a room in which active acoustics are permanently installed so that the recordings can be made over a longer period of time. In general, the concept of using active acoustics to change the characteristics of the room has been proven successful, as it allows the musicians to play and interpret music in an uncomplicated and natural way that does not distract from the actual musical performance due to complicated technical implementation. Also, for the recordings, it provides a possibility to quickly set up and dismantle reproducible setups without much effort, so that it is realistic to realize a comparable, but more extensive and thus probably more differentiated investigation with a larger pool of musicians. In a second step single acoustic parameters of the active acoustics could be adjusted to prove their specific influence on the interpretation.

 

7 REFERENCES

 

1. L. Beranek: Concert Halls and Opera Houses – Music, Acoustics and Architecture, Springer (2nd edition). (2004).

2. S. Bolzinger, O. Warusfel, E. Kahle: A study of the influence of room acoustics on piano performance, Journal de Physique IV, Colloque C5, 617–620. (1994).

3. J. Meyer: Acoustics and the Performance of Music, Chapter 8.4, Springer (5th edition). (2009).

4. K. Ueno, K. Kato, K. Kawai: Effect of Room Acoustics on Musicians' Performance. Part I: Experimental Investigation with a Conceptual Model, Acta Acustica united with Acustica, Vol. 96, 505–515. (2010).

5. K. Kato, K. Ueno, K. Kawai: Effect of Room Acoustics on Musicians’ Performance. Part II: Audio Analysis of the Variations in Performed Sound Signals, Acta Acustica united with Acustica, Vol. 101, 743–759. (2015).

6. Z. Schärer Kalkandjiev, S. Weinzierl: The Influence of Room Acoustics on Solo Music Performance: An Empirical Case Study, Acta Acustica united with Acustica, Vol. 99, No. 3, 433–441. (2013).

7. Z. Schärer Kalkandjiev, S. Weinzierl: Der Raum als Instrument. Zusammenhänge zwischen der Qualität der Akustik und der Spielweise von Interpreten, Studien und Materialien zur Musikwissenschaft, Band 78: Wahrnehmung – Erkenntnis – Vermittlung, Musikwissenschaftliche Brückenschläge, 167–177. (2013).

8. S. V. Amengual Garí: Investigations on the Influence of Acoustics on Live Music Performance Using Virtual Acoustic Methods, Ph.D. Dissertation, https://opus.hfm-detmold.de/frontdoor/deliver/index/docId/68/file/Dissertation_Amengual.pdf. (2017).

9. Z. Schärer Kalkandjiev, S. Weinzierl: The influence of room acoustics on solo music performance: An experimental study, Psychomusicology: Music, Mind, and Brain, 25(3), 195–207. (2015).

10. Z. Schärer Kalkandjiev: The Influence of Room Acoustics on Solo Music Performances. An Empirical Investigation, Logos. (2015).

11. T. Görne: Tontechnik, Hanser (4th edition), Chapter 9.2.5. (2015).

12. J. Meyer: Acoustics and the Performance of Music, Chapter 5.4, Springer (5th edition). (2009).

13. K. Ueno, H. Tachibana: Cognitive modeling of musician’s perception in concert halls, Acoustical Science and Technology 26, 156–161. (2005).