From classic to rap: Airborne transmission of different singing styles, with respect to risk assessment of a SARS-CoV-2 infection

3. Introduction

The coronavirus pandemic has had and still has a great impact on singing, especially while performing in groups, since the SARS-CoV-2 virus infection is transported through droplets from the respiratory system and can emit from the mouth opening to the surrounding air [1]. Hence, prevailing questions concern mostly distance regulations and the fundamental questions as to whether and how individual lessons or ensemble and choir rehearsals (in different settings, levels of profession, as well as amateur choirs) can take place. Since it is known that music making in general, and therefore singing, has an extended influence on a person’s well-being [2, 3, 4], these questions concern 4 million people in Germany who sing in a choir, of which 2.2 million amateur singers are registered in a choir union [5].

In the following, we need to know at what distance air movements are being produced while singing and how these air velocities can be evaluated in comparison to usual air movements, such as speaking or coughing, as well as considering usual room air velocities.

The measurements of our study took place in the concert hall of the Bamberg Symphony (see Appendix 1) at the beginning of May 2020, when singing in choirs indoors was mainly prohibited. Up to then, there were no general recommendations on choir rehearsals, singing in smaller ensembles or single teaching situations, due to the lack of well-founded scientific medical studies.

Considering up-to-date research, there are those looking into singers and/or wind instruments, focusing either on their respiratory airflows or measuring aerosol clouds.

The study by the Ludwig-Maximilians-Universität München and the University of Erlangen [6, 7] investigated different forms of speaking and singing with 10 professional singers of the Bavarian Radio Choir. They made respiratory flows visible by exhaling the smoke of e-cigarettes (not containing nicotine). Tests were conducted in a shaded room and documented the exhaled aerosol clouds with the help of a high-speed camera and laser light. Comparing different settings (singing text, speaking text and singing without text, once with soft and once with loud phonation), they found out that singing text and speaking text reaches comparable mean distances of up to 0.85m, while singing without text reached the lowest values of 0.63m. Even though the mean measurements stayed within a reach of 1m, some singers reached distances of up to 1.4m. These measurements were then compared to coughing, which turns out to reach even further, with a mean of 1.3m and a maximum of 1.9m. On the basis of their study, they announce a suggestion for distance regulations of at least 2m to the front.

Mürbe et al. [8] pre-published a study on the increase of aerosols during professional singing. Testing 8 professional singers (two female sopranos, two female altos, two male tenors and two male baritones) of the RIAS chamber choir Berlin, they measured particle emission rates with the help of a laser particle counter during breathing, reading, singing and holding long tones. They confirmed their assumptions of singing producing higher emission rates than speaking, with mean measurements of 4.71 – 84.75 P7s during speaking and 753.4 – 6093.14 P7s during singing. Women also produced higher particle emission rates than men, leading to the assumption that high voices produce higher sound pressure level than lower voices. On the basis of their findings, they assume that singing produces more emission rates of aerosols than speaking or breathing and that an increase of aerosol rates can be observed with an increasing sound pressure level during singing, especially while holding long tones.

In consideration of their study, Mürbe et al. [9], Kriegel & Hartmann [10], and Hartmann et al. [11] published various assessments on the risk of infection with the virus-loaded aerosols, while singing indoors during the SARS-CoV-2-pandemic. They assume that different styles of singing – e.g., singing vs. speaking – as well as different intensities of voice can lead to various sizes and density of droplets and aerosols [9] and that room situations of choir rehearsals have to be taken into account [10, 11].

Furthermore, Mürbe et al. found out that child voices emit fewer aerosols during singing than adults [12]. In this study they tested 8 children (four girls and four boys) of semiprofessional children’s choirs (Staats-und Domsingknaben Berlin and a girl’s choir of the Berlin Singakademie), who were all 13 years old (except one girl was 15 years old). The study was conducted the same way as their previous study on professional adult singers as mentioned above. Their mean measurements show emission rates of 16 – 267 P/s for speaking 141 – 1240 P/s for singing, and 683 – 4332 P/s for shouting.

In 2019, before the Coronavirus infection started, the study by Asadi et al. [13] observed different speaking situations regarding their particle rate of normal speech in correlation with loudness. They support the hypotheses that coughing and sneezing are emphasizing airborne infectious disease transmissions and point out that loudness of voice is related to the particle rate of aerosols, but also assume that further factors have to be considered.

Asahi et. al furthermore emphasize that the generation of aerosols increases with the amplitude of vocalization depending on loudness, but the language spoken (English, Spanish, Mandarin, or Arabic) showed no difference.

A study conducted by the University of the German Armed Forces in Munich, which was pre-published in May 2020 [14], analyzed larger droplets when singing and speaking, as well as the flow-related small droplets when singing and playing wind instruments. The study was conducted with a professional singer, two amateur choir singers, five professional musicians (clarinet, flute, oboe, bassoon, and trumpet), and an amateur brass player (trumpet, trombone, and euphonium). This study observed the motion of droplets and air leaving either mouth or outlet during exhalation, which was then illuminated with laser and recorded with a digital camera, producing a series of images that were quantitatively analyzed afterwards. The analysis points out that while singing, airflows at a distance of 0.5m were no longer detectable regardless of volume, pitch or whether the singer was a professional or an amateur. In conclusion, they recommend a minimal radial distance of 1.5m between singers and/or wind instrumentalists.

Another recent publication [15] conducted at the Bauhaus-Universität Weimar observed the spread of exhaled air while playing wind instruments and singing using the schlieren imaging with a schlieren mirror and the background oriented schlieren method (BOS) as a way to make respiratory air visible. Two professional singers (baritone and soprano) and eleven wind instruments (woodwinds: oboe, bassoon, Bb clarinet, bass clarinet, grand flute, piccolo, alto flute; brass: Bb trumpet, tenor trombone, French horn, F tuba) of the orchestra Thüringen Philharmonic Gotha, Eisenach, were positioned in front of the schlieren mirror while playing or singing. The findings show that the spreading range, as well as the angle at which the air escapes the mouth or outlet, varies strongly from instrument and player, depending on the structure of the instrument, the structure of the mouthpiece, the way an instrument is blown and individual blowing or breathing capacities.

Based on our own studies, we updated our risk assessment for the fifth time, which was first published in April 2020 [17]. Herein we suggest a 2m radial distance between singers, as well as consideration of other factors for safe singing settings, such as social behavior and frequent ventilation to refresh CO□ emissions [18]. Our study on CO□ measurements in instrumental and vocal closed room settings showed that singers have lower emission rates than speaking or listening, leading to recommendations on frequent ventilation breaks of 5 to 10 minutes.

On behalf of this background, our study aims to provide evidence to the field of singing and music making, by providing funded data on velocity, direction and distance of respiratory air while singing different styles and pitches in comparison to speaking, as well as usual room air velocities.

4. Materials and Methods

5. Results

6. Discussion

This study observed three professional singers: a classically trained male baritone singer, a classically trained female soprano singer and a non-classically trained female singer, who all performed 15 different singing scenarios: different singing styles, speaking and blowing. Since the measurements consider various singing styles as well as comparative measurements, it has a high external validity. The findings are very representative of a high level of professional singing, with limited transferability to the amateur sector.

In order to evaluate the results of our study, in relation to recently published articles on the generation of airflows or aerosol clouds while singing, our main finding reveals that directed blowing – which is comparable to coughing – reaches much higher measurements of air velocities than singing, with the highest value of 0.7 m/s at the 1m sensor (performed by the baritone singer). At 1.5m the value dropped to 0.34 m/s and at 2m still reached a measurement of 0.22 m/s – which is a value that lies above the usual room air velocity of 0.1 – 0.15 m/s. Therefore, the risk of infection is higher while coughing rather than singing. This is a finding that is also supported by Asadi et al. [13] and Echternach et al. [6].

When comparing the measurements for singing and speaking, they show the same maximal measurements of air velocities of up to 0.17 m/s at a distance of 1m, dropping to 0.12 – 0.13 m/s at the 1.5m sensor, and staying below the usual room air velocity of 0.1m/s at the 2m sensor. In the following, for both settings of speaking and singing, air movements were no longer measurable at a distance of 2m and can therefore be considered as insignificant airflows [23].

Hence, significant airflows were measured only at 1m, with low measurements at the 1.5m sensor, which decline within or below the usual room air velocity of 0.1 – 0.15 m/s, a finding supported by Becher et al. [15], Kähler & Hain [14], Mürbe et al. [8, 12] and Echternach et al. [6].

With regard to the relation between loudness of voice and the dispersion of aerosols, our study did not find any relation, whereas other studies did, such as Asadi et al. [13] or Mürbe et al. [8, 12].

Taking into account different styles of singing, we observed that they make almost no difference at all in the generation of increasing airflows. A significant difference between blowing and singing or speaking was observed, but not by use of different singing styles. In this case, we compared mainly the highest merits of measurements of a sequence.

All qualitative observations were made close to the body of the singing person, which lead to the presumption of low measurements at the sensors, especially in distances beyond 1.5m. Quantitative measurements supported these observations, showing that respiratory air movements while singing are not dependable on volume, pitch or style of singing, but rather on individual singing habits and possibilities, which was also observed in the study of Becher et al. [15].

It is remarkable and interesting that we found differences in our assumptions after seeing the qualitative observations and having the merits from the quantitative measurements. On the basis of the qualitative observations only, it was assumed that plosive tones or rap singing would produce higher air velocities than other singing styles. The measured results showed differently: We could not find a difference if a singer is singing a song, speaking or using plosive tones – all measurements stay at the same level for all three test persons. Other studies found out otherwise, observing differences between different speaking and singing, especially while holding long tones [8], or a difference between singing with or without text, but no difference between singing or speaking text [6].

Comparing the findings on respiratory air movements while singing with our own recent study on woodwinds, it was found out that singers reach higher air velocities than woodwinds [16]. These measurements of airflows for singers stay within a reach of 1.5m, which were only surpassed during directed blowing. The highest air movement measurements for woodwinds were produced by the oboe (with a maximal value of 0.15 m/s at 1m) and the tuba (with a maximal value of 0.13 m/s at 1m), while singers reached 0.17 m/s at 1m while performing an excerpt of a music piece. Particular to instrument playing was that air-jet woodwinds, such as alto flute or piccolo, also produced side air movements of up to 0.15 m/s (alto) and 0.13 m/s (piccolo). For the other wind instruments room air movements above 0.1 m/s were no longer measurable at 1.5m while playing. They only surpassed usual room air velocities when blowing out or warming up (reaching maximal values of 0.5 m/s at 1m).

With regard to our risk assessment [17], we suggest a radial distance regulation of 2m for singers, especially when singing in groups. Only then, droplet infection due to direct contact can be excluded. This recommendation also supports the fact that strong airflows can transport heavy droplets even further than 1.5m. On the basis of our findings, though, measurements of airflows beyond 1.5m while singing were not visible. We were furthermore unable to measure jet-stream droplet effects of light droplets, which would have been visible in the artificial fog. Furthermore, we suggest frequent ventilation breaks, as mentioned in our study on CO□ measurements [18], to prevent a coronavirus infection while performing musical activities.

Aside from our findings and risk assessments, some limitations of our study have to be taken into account.

First, the study was conducted with highly professional singers, and the results can therefore not automatically be transferred to other settings or amateur singers.

Second, the test results are rather specific. We took into account different singing styles and three different singers from two different musical backgrounds. Therefore, the findings are very representative for professional classical singers as well as modern rock/pop singers, with restricted transferability to the amateur music sector and large choir groups. Since the test singers have been part of the study and know their voices very well, relevant lung capacity or ways of singing were identified for every test person. This seems to be one of the strengths of our study.

Further limitations concern the fact that the measurements were conducted with three singers only, while being sensitive to individual differences of singing, lung volume, etc. They were also performed only once for every singer, whereas more repetitions of the same sequence or analyzing more persons would have given more information on reproducibility of the test setting.

Another limitation of the measurements is the fact that air velocity measurements are very sensitive to surrounding movements, with the waving of a hand already influencing the measurements at the sensor.

7. Conclusion

The test results show that singing is less dangerous than coughing and comparably produces the same air velocity as speaking. In our study, results do not depend on volume, pitch or the style of singing. In this spectrum, the highest measurement of singing lies at 0.17m/s at a distance of 1m to the singer. At 2m, measurements over the usual room air velocity of 0.1m/s were no longer visible, while frontal and direct blowing, which is comparable to coughing, rise up to 0.7 m/s at the most at 1m.

Since no respiratory air movements – of any singing style – were measured at the 2m sensor, we find distance regulations of 2m to the front of the singing person to be maintainable.

In order to maintain a responsible risk assessment, we find it crucial that besides a regulated distance setup (e.g., small or large ensemble), constant fresh air ventilation and social behavior should also be taken into account. Starting from April 2020, the Freiburg Institute of Musicians’ Medicine constantly updated an official paper on “risk assessment of a coronavirus infection in the field of music” (updated May 19, July 1, July 17, and December 14, 2020), stating the findings of our various studies publicly [17] and establishing a permanent consultation on questions concerning the relationship between the coronavirus and music making.