Applied Acoustics 178 (2021) 108051

Contents lists available at ScienceDirect

Applied Acoustics
journal homepage: www.elsevier.com/locate/apacoust

How much COVID-19 face protections influence speech intelligibility

in classrooms?
Marco Caniato ⇑, Arianna Marzi, Andrea Gasparella
Free University of Bozen Bolzano, Italy

a r t i c l e i n f o a b s t r a c t

Article history: The ongoing pandemic caused by the COVID-19 virus is challenging many aspects of daily life. Several
Received 14 December 2020 personal protective devices have become essential in our lives. Face protections are mostly used in order
Received in revised form 9 March 2021 to stop the air aerosol coming out of our mouths. Nevertheless, this fact may also have a negative effect
Accepted 11 March 2021
on speech transmission both in outdoor and indoor spaces. After a severe lockdown, classes have now
Available online 23 March 2021
started again. The adoption of face protection by teachers is either recommended or mandatory even
though this is affecting speech intelligibility and thus students’ comprehension. This study aims to under-
Keywords:
stand how protections may affect the speech transmission in classrooms and how this could be influ-
COVID-19
Pandemic
enced by the several typologies of face protections. An experimental campaign was conducted in a
Face mask classroom in two different reverberant conditions, measuring and comparing the variation in speech
Surgical transmission and sound pressure level at different receiver positions. Furthermore, a microphone array
Acoustic was used to investigate the distribution of the indoor sound field, depending on the sound source.
Hearing Results clearly show how different types of personal protection equipment do affect speech transmission
and sound pressure level especially at mid-high frequency and that the source emission lobes vary when
wearing certain types of personal devices.
Ó 2021 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://
creativecommons.org/licenses/by/4.0/).

1. Introduction
At present, the world is dealing with the SARS-CoV-2 or COVID-
19 pandemic. Several precautions were taken to limit the spread of
the virus. Social distances and lockdown were imposed in all coun-
tries, leading to a decrease in contagious behaviour and situations.
When these actions were mitigated, governments made individu-
als use protection at least in indoor environments. Thus, face
masks, shields and respirators were adopted everywhere, becom-
ing a necessary element of everybody life. However, personal facial
protections are already widely used in medical care and in hospi-
tals to prevent healthcare workers from virus infections by aerosol
particles [1]. They have also demonstrated to be efficient personal
devices that slow down the spread of the Corona virus [2]. How-
ever, there are several different types of protective equipment
available on the market: (i) surgical masks, (ii) cloth or community
masks [3] or respirators (also known as Filtering Face Piece) and
(iii) face shields [4].This variety of protections not only has an
impact on the filtering of aerosol particles [5], but also on commu-
nication capabilities. Moreover, face masks are a visible obstacle in
⇑ Corresponding author at: Piazza Università, 39100 Bozen, Italy.
E-mail addresses: mcaniato@unibz.it (M. Caniato), amarzi@unibz.it (A. Marzi),
andrea.gasparella@unibz.it (A. Gasparella).
body language, impacting: (i) verbal communication [6], (ii) peo-
ple’s emotional expressivity, the signs of understanding [7], (iii)
lip reading [8], which is a very useful aid in ordinary communica-
tion among people and extremely useful for people with hearing
loss and finally (iv) voice [9].
For the above-mentioned reasons, it is therefore essential to
study how face protections may influence noise source in the built
learning environment. The adoption of these devices is certainly
needed for the pandemic [10], but they could negatively influence
communication. Indeed, several studies have already demon-
strated the impact of these devices on speech. Aranaz Andrés
et al. [11], in their study on the secondary effects of face masks,
highlighted the difficulties related to speech understanding, when
speakers are wearing face protection. Hulzen and Fabry [8] inves-
tigated their effects on COVID-19 infection protection and the dis-
comfort caused to individuals with hearing loss due to the use of
face masks. Accordingly, face protections pose two problems for
these people: they cannot receive any clues from lip reading [12]
and furthermore, other people’s voices may be attenuated or in
some way distorted.
Many available studies focus on communication difficulties in
healthcare environments. Wittum et al. [13] investigated the
side-effects of wearing surgical masks in hospitals. Goldin et al.
[14] compared, in a controlled environment, a few face masks

https://doi.org/10.1016/j.apacoust.2021.108051
0003-682X/Ó 2021 The Authors. Published by Elsevier Ltd.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
M. Caniato, A. Marzi and A. Gasparella
applied to a white noise simulator through an artificial mouth.
They compared the difference in sound pressure levels between
all the masks offered by their healthcare providers, highlighting
that each type of device essentially reduced the speech transmis-
sion in this environment. Other studies have demonstrated the
impact of face masks on communication and on speech. Corey
et al. [15] measured the sound reduction caused by several masks,
focusing on sound pressure levels in a laboratory with sound-
absorbing walls. Conversely, Palmiero et al. [16] studied the Speech
Transmission Index of three masks to understand how these may
compromise sound transmission.
During the pandemic, in several countries the use of masks
inside educational environments has either become mandatory
for teachers and/or for students [17,18], or recommended [19], in
order to protect teachers, students and their families [20].
From this perspective, the investigation of how these protec-
tions may influence learning and communication performance
assumes a key role. There are various types of educational environ-
ments. In some countries there are standards for acoustic condi-
tions in the classes that have thresholds for reverberation times,
noise levels and sound pressure levels at receivers’ positions [21].
However, generally, the classroom can have a high reverberation
time (reverberant conditions) [22] or may feature sound absorbing
material to reduce the reverberation [23,24].
Moreover, the educational environment is strongly influenced
by the indoor acoustic conditions [25]. Indeed, in classrooms more
than one factor is generally taken into account: (i) background
noise [26], (ii) acoustic field [27], (iii) effects of noise on the
speaker/teacher [28] and (iv) effects on the listener/students
[29]. Skarlatos and Manatakis [30] found that reverberation time
is one of the main causes of high noise levels in classrooms. As also
mentioned by Hodgson et al. [31], this parameter is of paramount
importance and it should be taken into account when characteriz-
ing indoor spaces. It refers to the reflection of sound waves [32]
and it has been shown to have a strong impact on classroom
soundscape [33] and on students’ achievements [34]. Indeed, con-
trol of the indoor sound field improves pupils’ speech comprehen-
sion [35] and it has been shown that a good acoustic environment
helps to increase voice and words understanding [36]. Generally, in
order to assess the acoustic quality of a room, specific objective
parameters are used, coupled with reverberation time. When
focusing on rooms dedicated to speech, clarity is one of the most
utilized parameters [37]. Clarity considers reverberation time and
acoustic strength as a single index and thus it is very useful in
determining how the indoor sound field varies when moving from
one student’s desk to another [38]. However, these parameters are
not affected by possible source signal amplitude variations. Thus,
they can effectivly assess indoor sound field, but they cannot be
used to investigate source-receiver sound wave transmission.
Accordingly, Zannin and Zwirtis [39] measured the sound pres-
sure levels in three Brazilian classes and found that this parameter
is not only related to the quality of verbal communication, but also
to the listening effort, which can impact students’ activities. Astolfi
et al. [40] in their study highlighted how (i) sound pressure levels
measured inside the classroom, (i) reverberation time and (iii)
numerical models should be integrated to have a systematic proce-
dure to approach this complex indoor environment. Within the
learning environment, Prodi et al. [41] found that the Speech
Transmission Index (STI) is also a relevant parameter strongly
related to Clarity and the measured sound pressure levels at the
receivers’ positions. Seetha et al. [42] reported in their research
that the STI measurement coupled with the analysis of the sound
pressure levels is enough to have a complete overview of the sound
field in a classroom. Peng et al. studied STI in 28 Chinese schools,
finding that the speech intelligibility score is a very important
parameter for all ages. Radosz [43] illustrated a method to assess
2
Applied Acoustics 178 (2021) 108051
the acoustic quality of classrooms including all the factors men-
tioned above with a unique global index.
The importance of students’/listeners’ distance from the source/
teacher has also been assessed. Bilzi et al. [44] highlighted how in
classrooms STI could significantly vary from one position to
another, while Stewart and Cabrera [45] found that the directivity
of the source strongly influences the STI results in noise-free con-
ditions. Thus, reverberation time clarity and STI are the parameters
most often used to characterize indoor sound field, while STI and
sound pressure levels at receivers’ positions are needed to investi-
gate the quality of the transmitted speech.
Among all these medical and acoustic studies, there is a lack of
focus on the impact of the face protections within educational
environments. Students in classrooms are careful listeners
[46,47]. Wearing masks may impact the communication between
teachers and students, affecting linguistic and non-verbal informa-
tion [48]. Transparent masks or shields may represent one solution
as they may help by making lip reading possible and helping to
overcome speech degradation [49]. However, to the authors’
knowledge, no studies have been conducted on these issues.
For these reasons, the aim of this research is to understand to
what extent face protections influence (i) speech intelligibility,
(ii) indoor sound pressure level distribution and (iii) source direc-
tivity within a classroom. Another aim of this research is to under-
stand if the impact of these devices varies, according to diverse
indoor sound field conditions.
2. Materials and methods
This study was conducted in the Free University of Bozen living
lab, which provides reconfigurable educational environments that
are used for everyday activities. The facility includes a classroom,
wherein it is possible to change indoor sound absorption and thus
to study diverse acoustic conditions.
Two different scenarios were built:

Scenario A (Higher reverberant room), where a reverberant
classroom was recreated, emulating a traditional school
environment;

Scenario B (Lower reverberant room), where 16-square-meter
sound absorbing panels were installed on the room ceiling.
Their sound absorption coefficient is reported in Fig. 1. Panels
were composed of 5 cm polyester fibre panels with a porosity
of 0.99. The air gap between them and the room ceiling is
10 cm.
Indoor sound field acoustic characterizations were performed in
accordance with the ISO 3382 standard [50–52] using an omnidi-
rectional (dodecahedral) sound source and a logarithmic sine
Removable panels sound absorpon coefficient
1
sound absorpon coefficient [-]
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Frequency [Hz]
Fig. 1. Panels sound absorption coefficient.
M. Caniato, A. Marzi and A. Gasparella
Fig. 2. hanging microphones and dodecahedral source positioning.
sweep as a signal, as depicted in Fig. 2. One excitation per one
receiver position was played, recorded and then post produced
by means of a convolution, in order to obtain the impulse response.
The source position was chosen very close to the teacher’s desk.
The positions of the receivers were uniformly distributed along
the rows of desks of the existing students. Reverberation time
T30 and clarity C50 were derived from measurements and used to
characterize different scenarios.
In order to understand how face protections may influence
sound source (teachers), speech intelligibility tests were carried
out using a repeatable and robust procedure so as to ensure reli-
able results. In particular, a constant and directive sound source
is needed to ensure comparable values. This procedure determines
the Speech Transmission Index (STI). It assesses the quality of the
received speech in different student positions and it is normally
used to determine whether a classroom is suitable for teaching
and learning or not. STI evaluation is based on IEC 60268-16 [53]
which provides a gender specific octave band weighting and
redundancy factors. Gender-related factors are expressed differ-
ently due to signal spectra and different weighting factors.
The STI measurement is strongly influenced by several factors
related to the acoustic conditions of the classroom [54]: the
Fig. 3. used mouth and nose adapters.
3
Applied Acoustics 178 (2021) 108051
reverberation time, background noise, clarity and sound pressure
levels. Evaluations connected to speech were assessed using
speech intelligibility tests [19], developed using a directive MLS-
equipped noise source, positioned on the teacher’s desk and recei-
vers located at the students’ positions [55]. Unlike clarity and
reverberation time, which depend exclusively on (i) the classroom
volume, (ii) the internal reverberation time and (iii) the source-
receiver position, the STI also relies on the sound source frequency
level. If this is altered, the final result varies.
In terms of background noise, it was regularly checked so as not
to disturb or influence the acoustic tests. In this way, especially for
STI measurements, it was possible to consider the uncertainty to be
lower than a JND [56]
Since the face protections are not normally attached to real
mouths, special accessories were plugged on a directive speaker
surface in order to maintain the masks’ membrane at a distance
of about 4 mm to 7 mm from the directive speaker in order to sim-
ulate a mouth-nose appearance and profile (Fig. 3).
Protections were then mounted on the directive speaker, using
this configuration. In order to analyze as many options as possible
within the two described scenarios A and B, ten kinds of individual
face protections were tested. They were mounted on the directive
sound source, covering the noise emission so as to simulate real
mouth and nose overlapping. The devices used are listed in Table 1
and examples of applications are reported in Fig. 4. Complete iden-
tification of the face masks under consideration and transparent
shields are reported in the Appendix tables (see Fig. 5).
Tests were performed using a B&K 4720 directive noise source
and a B&K 2270 sound level meter, connected to a sound card con-
trolled by B&K Dirac 6.0 software. The receiving points were placed
at 1.4 m from the ground in six representative positions, according
to the class layout (Fig. 6). The room is box-shaped, 7.3 m  7.6
m  3.6 m, for a volume of 196 m3. The source was placed at
1.5 m from the ground and near the teacher’s desk. As consistent
with COVID guidelines, the source (teacher) and the positions of
the receivers (students) are not expected to change [58]. For every
scenario, firstly a traditional STI measurement was performed, i.e.
without any face protection (case 0), as a starting point for the
comparison. For the sake of simplicity, only frequency average
results will be presented. These are calculated by taking into con-
sideration the octave band measured valued of 500 Hz, 1000 Hz
and 2000 Hz. An arithmetic average is then performed, in order
to obtain an average STI result.
Reverberation time, clarity and speech transmission index vari-
ations are evaluated using the Just Noticeable Difference (JND)
approach [50]. Seraphim [59] stated that a reverberation time vari-
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

Table 1
List and description of all masks and combination.

ID Name of the mask Description Available Filter type Certified

number in PPE
pharmacy
1 Surgical Surgical masks Yes – No
2 Surgical Masks featuring stiff Surgical mask with a transparent insertion dedicated to lip No – No
transparent insertion reading. Transparent insertion is made of Polyethylene
Terephthalate Glycol
3 Surgical Masks featuring Surgical mask with a transparent insertion dedicated to lip No – No
flexible transparent insertion reading. Transparent insertion is made of Transparent
acrylic sheet
4a Community Mask - no filter Two layers of cotton fabric, featuring a possibility of filter No – No
[57] insertion. No filter
4b Community Mask - with filter Two layers of cotton fabric, featuring a possibility of filter No Non-woven polypropylene (PP) and No
[57] insertion. Paper filter paper filter
5 Hand-made face Mask, kid Two layers of cotton fabric, featuring a possibility of filter No – No
size insertion
6 FFP2 safety mask Filtering face mask made of polypropylene melt blown Yes Removes at least 94% of all particles Yes
cloth starting from 0.3 mm in diameter
7 NCN MR2 Certified organic cotton and treated with exclusive Carbon No NCN materials with high anti- Yes
Nanoclusters. particle, antimicrobial and
bacteriostatic properties
8 Face shield Polyethylene Terephthalate Glycol Yes – No
9 Face shield coupled with Face shield made of Polyethylene Terephthalate Glycol, or Yes (face – No
surgical mask PETG combined with a surgical mask. shield
Yes (face
mask)
10 Face shield coupled with Face shield made of Polyethylene Terephthalate Glycol Yes (face – No
community Mask – with combined with mask made with two layers of cotton fabric. shield)
paper filter
No (face Paper filter
mask)

Fig. 4. examples of complete tests. Left: cotton mask. Right: transparent shield.

Fig. 6. Classroom layout: positions of the noise source and receivers for STI
Fig. 5. Face protections investigated useful for COVID 19 spread prevention. measurements. Locations of students are depicted by chair occupancy.

4
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

ation is perceived when its variation is at about 5%. Karjalaine and
Jarvelainen [60] reported that a 10% variation is more accurate
than 5%. For this reason, in this paper a variation of 10% is used
to consider a reverberation time variation to be significant. For
clarity and STI, a variation of 1 dB for clarity and 0.03 for STI are
to be classified as minimum JNDs, according to Bradley et al. [61].
In order to better understand indoor sound field variations, syn-
chronous sound pressure level measurements were developed by
means of an omnidirectional microphones model ECM ½” 999
hung on the roof and controlled by a Zoom F8 sound card. The posi-
tions are depicted in Fig. 6. They were located at a height of 2 m
(Fig. 2).
Furthermore, since source directivity may be affected by mouth
obstruction, acoustic camera high frequency analyses were carried
out using a planar 40 cm  40 cm microphone array, equipped
with 40 MEMS noise sensors coupled with an A/D converter. Every
device features a linear frequency response from 60 Hz to
15000 Hz. Software analysis was performed using a Robust Asymp-
totic Functional Beamforming algorithm, providing high quality 5
megapixels real time images and videos. The acoustic camera
was situated in receiver 4 position (Fig. 6), in order to better mon-
itor the phenomenon.
3. Results and discussion
In terms of the acoustic characterization of the room scenarios,
Tables 2 and 3 report reverberation time and clarity for mid-high
frequencies (500 Hz – 2000 Hz) and STI results for case 0 (no
device).
It can be seen that in Scenario A (higher reverberant room),
most of the parameters present values which are usually conducive
to a good indoor learning environment [62] at most considered fre-
quencies. Accordingly, reverberation time is over 1 s in all fre-
quency bands in all positions, while clarity results are close to
zero or below in almost all cases. As for the STI results, it is clear
that Scenario A provides poor speech intelligibility [63] in all posi-
tions, regardless of the teacher gender, except in position 1 (very
close to the source) and in position 3 (in front of the source).
In terms of Scenario B (lower reverberant room), it can be seen
that indoor sound field quality greatly improved. Reverberation
time is always lower than 0.75 s, highlighting a variation of up to
4 JNDs. Clarity is always higher than 2 dB, except in position 2
and 4 at 500 Hz, where, however, the values are always higher than
zero. This shows a positive variation up to 4 JNDs. The value of the
STI also improved, gaining up to 3 JND per position per device type.
As an overall result, it is possible to state that the indoor sound
field significantly varied its conditions and thus STI measurements
using face masks in scenario A and scenario B start from clearly dif-
ferent indoor conditions.
3.1. STI results
The results of STI measurements performed using individual
protection were grouped based on the receiver position and cata-
logued based on the wearable protective equipment. For the sake
of brevity, all tables are reported in appendix A. As a reference,
the case without any face protections (case 0 – no device) was
used, which as expected presents the best STI values. As an overall
result (Table 4), it can be highlighted that many individual protec-
tions highly limit sound wave propagation. Considering the maxi-
mum JND results, up to 2 JNDs were found in all positions.
Maximum values show how STI male values are 2 to 3 JNDs for sce-
nario A and 2 to 5 JNDs for scenario B. For STI female values are
lower: 1 to 2 JNDs, for the first case, while 1 to 3 for the second
case.
In Fig. 7, a representative background noise level is reported.
This was measured during the test. As it can be seen, levels are very
low in every frequency. Since the noise source level has to be 60 dB
(A) at 1 m from the noise source, it can be concluded that back-
ground noise cannot influence the final results.
Another general outcome is that differences are higher in sce-
nario B (lower reverberant room), rather than scenario A (higher
reverberant room). It is also evident how the STI male in both cases
suffers more compared to the female STI. Positions 3 and 6 are
where maximum variation are verified for both scenarios, followed
by Position 2 and Position 1, which have a similar variation of STI
in both scenarios. In addition, as expected, the face shield coupled
with the community mask featuring a paper sheet is the worst

Table 2
Reverberation time and clarity frequency results.

Position ID Scenario A Scenario B

T30 C50 T30 C50
500 Hz 1000 Hz 2000 Hz 500 Hz 1000 Hz 2000 Hz 500 Hz 1000 Hz 2000 Hz 500 Hz 1000 Hz 2000 Hz
1 1.22 1.05 1.05 0.42 0.25 0.48 0.74 0.70 0.72 4.85 3.96 3.89
2 1.16 1.08 1.06 1.28 0.41 1.12 0.69 0.69 0.70 0.25 3.12 1.86
3 1.17 1.06 1.07 2.37 0.16 0.81 0.72 0.68 0.73 2.37 3.77 3.64
4 1.28 1.10 1.07 1.11 0.31 0.55 0.66 0.67 0.73 0.48 2.18 2.89
5 1.12 1.08 1.04 0.67 1.25 0.02 0.69 0.69 0.72 3.29 2.14 2.65
6 1.21 1.11 1.04 1.60 0.47 0.84 0.73 0.7 0.69 4.02 2.21 2.46

Table 3
– Speech transmission index results for case 0 (no device).

Scenario A Scenario B
STI Male STI Female STI Male STI Female
1 0.51 0.50 0.56 0.56
2 0.48 0.48 0.54 0.54
3 0.51 0.50 0.60 0.59
4 0.48 0.48 0.52 0.52
5 0.48 0.48 0.50 0.51
6 0.48 0.48 0.50 0.51

5
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

Table 4
– maximum and minimum STI JNDs related to different face protections.

Scenario A Scenario B
Position STI male max JNDs STI Female max JNDs STI male max JNDs STI Female max JNDs
1 3 ID 10 1 ID 10 4 ID 10 3 ID 10
2 3 ID 6 2 ID 6 4 ID 10 2 ID 10
3 3 ID10 2 ID 17 5 ID 10 3 ID 10
4 2 ID 10 1 ID 10 4 ID 8 2 ID 8
5 2 ID 4b 1 ID 4b 2 ID 4b 1 ID 4b
6 3 ID 7 2 ID 7 4 ID 4b 3 ID 4b
STI male min JNDs STI female min JNDs STI male min JNDs STI Female min JNDs
1 1 ID5 0 ID5 0 ID1 0 ID1
2 0 ID1 0 ID1 1 ID1 0 ID1
3 1 ID1 0 ID1 2 ID1 1 ID1
4 0 ID5 0 ID5 1 ID5 0 ID5
5 1 ID1 0 ID1 1 ID8 0 ID8
6 1 ID8 1 ID8 0 ID8 0 ID8

measured background noise level to the face shield coupled with the community mask (ID 10), fol-
lowed by the community mask with a paper filter (ID 4b), NCN
30
MR2 mask (ID 7), the face shield (ID 8) and the FFP2 safety mask
sound pressure level (dB(A))

25 (ID 6). This is attributable to the fact that they are all medical or
high efficiency protection devices.
20
Referring to minimum differences in JNDs, it can be seen that
15 surgical masks (ID 1) are the best masks, followed by kids’ masks
(ID 5). Values show that very few JNDs are caused by surgical
10 masks in most receivers’ positions in both scenarios and for both
5
male and female sources.

0
3.2. Sound pressure level results

Fig. 7. Representative background noise level during the measurements.
combination, followed by the community mask featuring a paper
filter (ID 4b). The maximum JNDs assessed are mostly referable
In order to understand how the indoor acoustic field is affected
by source protection, sound pressure level measurements were
taken. For the sake of brevity, results for the receivers 2, 3 and 6
are shown in Figs. 8 and 9, reporting on the x-axis the difference
between the reference case (ID 0 – no face protection) and the

Fig. 8. SPL relative differences frequency trends for face protections which least affect indoor sound field. Positions 2, 3 and 6.

6
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

Fig. 9. SPL relative differences frequency trends for face protections which most affect indoor sound field. Positions 2, 3 and 6.

investigated device (e.g. ID 0 minus ID 1). In order to present an
overview of both the least and the most impacting face protection,
the results are listed in Table 5. In this table, the sum of the sound
pressure levels at voice frequency (630–5000 Hz) Lp and the differ-
ence between the reference case (ID 0 – no face protection) and
each device D are reported.
It is evident that surgical masks, even those featuring a trans-
parent window, do not significantly modify sound pressure levels
at the receivers. Other types of masks provide significant devia-
tions of up to 2.7 dB. Face shields clearly alter the sound wave
transmission. Negative differences mean that visors reflect a signif-
icant part of the sound waves to the upper part of the room, thus
considerably affecting the source-receiver propagation path.
On the frequency domain, some results offer interesting
insights. In Figs. 8 and 9, the results of respectively the least and
the most influencing devices are reported. Accordingly, it can be
seen that all individual protections provide differences at mid-
high frequencies. Accordingly, volcano-shaped trends at 2000 Hz
can easily be identified in both scenarios. In the cases reported in
Fig. 8, levels can vary from 5 to 10 dB in both scenarios, while
the ones included in Fig. 9 fluctuate from 8 dB to 11 dB in scenario
A and from 9 dB to 14 dB in scenario B. Thus, these protections may
modify the original emitted signal in a range where human voices
are very active. A reduction of up to 10 dB in both reverberant and
absorbing rooms was found, thus implying that sound pressure
reductions do not depend on indoor sound field conditions.
On higher frequencies, some differences are still present. They
linearly decrease to 4000 Hz, where they stop and in some cases
start to slightly increase again. Accordingly, both scenarios present
the same overall trends. The face shields (ID 8 and ID 10) offer a
positive effect at low and high frequency, but when it is coupled
with a mask, they negatively influence soundwave propagation
only at high frequency.
In this light, it is interesting to present tridimensional relative
difference trends of sound pressure levels related to the single
most influencing face protection in order to understand if there
are some significant differences among the diverse devices.
Fig. 10 shows scenario A, while Fig. 11 shows scenario B. For the

Table 5
Sound pressure level of voice frequency (630–5000 Hz) Lp and relative case difference D. Grey: minimum differences; plain: medium differences; bold maximum differences.

Position n.2 Position n.3 Position n.6

Scenario A Scenario B Scenario A Scenario B Scenario A Scenario B
ID Lp (dB) D (dB) Lp (dB) D (dB) Lp (dB) D (dB) Lp (dB) D (dB) Lp (dB) D (dB) Lp (dB) D (dB)
0 56.8 – 57.6 – 58.4 – 56.7 – 55.0 – 52.2 –
1 55.9 1.0 56.9 0.7 57.9 0.5 55.9 0.8 54.4 0.6 52.2 0.1
2 56.0 0.9 56.6 1.0 57.9 0.5 55.9 0.8 54.6 0.4 51.8 0.4
3 55.7 1.1 56.7 0.9 58.1 0.3 55.9 0.7 55.0 0.0 51.9 0.3
4a 54.7 2.1 56.4 1.1 56.9 1.5 54.5 2.2 53.5 1.5 50.9 1.3
4b 54.9 1.9 56.0 1.6 57.4 1.0 53.9 2.7 53.8 1.2 50.6 1.6
5 55.1 1.7 56.5 1.1 56.7 1.7 54.6 2.1 53.5 1.5 50.8 1.5
6 55.1 1.7 56.8 0.8 57.3 1.2 54.7 2.0 54.1 0.9 51.2 1.0
7 55.3 1.5 56.6 1.0 57.6 0.8 54.8 1.9 53.9 1.1 51.3 0.9
8 60.7 3.9 60.6 3.0 62.8 4.4 61.0 4.3 59.8 4.8 57.3 5.1
9 60.1 3.3 59.6 2.0 62.5 4.1 60.2 3.5 58.9 3.9 56.8 4.5
10 60.2 3.4 59.7 2.1 62.2 3.8 59.6 2.9 59.3 4.3 56.5 4.3

7
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

11.2
13.4 ID 4b
10.8 ID 6
13
12.6 10.4

12.2 10
11.8 9.6
11.4 9.2
11 8.8
10.6
8.4
10.2
8
9.8
7.6

10.8
14
10.4
13.6 ID 7 ID 10
10
13.2
9.6
12.8
9.2
12.4
8.8
12
8.4
11.6 8
11.2 7.6
10.8 7.2
10.4 6.8

Fig. 10. 3D SPL relative difference distribution for the 5 most impacting face protections at 2000 Hz. Scenario A.

first scenario, it can be deduced that all face protections provide
similar 3D distribution. Differences are present in all receiver posi-
tions and always show significant values, starting from the closest
to the most distant positions. Specifically, it can be highlighted that
position 6, which is also the furthest from the noise source, always
shows the highest value. Conversely, locations 1, 2 and 3 very often
provide the same results. As an overall result, we can deduce that
the majority of the face protections, which clearly influence sound
propagation in a reverberant classroom, will provide almost the
same sound field variation.

13.8
13.6 ID 4b 8.2 ID 6
13.4 8
13.2
13 7.8
12.8
12.6 7.6
12.4
7.4
12.2
12 7.2
11.8
11.6 7
11.4
11.2 6.8
11 6.6
10.8
10.6 6.4

11.6 ID 10
11.2
11.4 11
ID 7 10.8
11.2
10.6
11 10.4
10.2
10.8 10
10.6 9.8
9.6
10.4 9.4
10.2 9.2
9
10 8.8
9.8 8.6
8.4
9.6 8.2

Fig. 11. 3D SPL relative difference distribution for the 5 most impacting face protections at 2000 Hz. Scenario B.

8
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

In the case of a sound absorbing environment, differences are

Mask ID 0 Mask ID 1
always distributed using the same pattern. The worst location (ex-
cept mask ID 10) is position 3, where the highest difference can be
found. In the first row of desks, values are often the lowest ones
and in position 3 results are the lowest for mask ID 4b and 6. Fur-
thermore, in scenario B, we can conclude that all face protections,
which regularly affect noise propagation, present common sound
field differences.
In terms of sound pressure level, as a global result it can be
determined that differences are significantly high in both scenarios
and they provide the same difference ranges. The asymmetry could Mask ID 2 Mask ID 3
be caused by the sound absorbing panels distribution on the ceil-
ing. This implies that there is no significant difference provided
by the indoor environment conditions, but the sound limitations
are intrinsically related to the face mask. Another paramount result
is that face protections mostly act in a difference range of 8–14 dB
at 2000 Hz. This means that 8 dB of source reduction is granted
even if a receiver (student) stands close to the sound source (tea-
cher), regardless of the position and of the indoor acoustic condi-
tions. Furthermore, the last receivers’ row, which is the furthest
away from the source (teacher), present the worst difference val- Mask ID 4a Mask ID 4b
ues, up to 14 dB. This means that wearing one of these face protec-
tions may reduce the sound wave emitted by the speaker by up to
14 dB.

3.3. Acoustic camera results

Acoustic camera results are reported in Fig. 12 for scenario A

and in Fig. 13 for scenario B, both for 2000 Hz, for the sake of brev-
ity. When comparing the two different indoor acoustic fields, as an
Mask ID 5
overall result it is evident that in a more reverberant room the Mask ID 6
source emission presents a wider emission lobe. The application
of face masks modifies the received sound pressure levels but does
not significantly variate the lobes’ dimensions. On the other hand,
face shields do change emission patterns, splitting them into two
distinct parts.
This effect is caused by the sudden reflection of the direct sound
wave against the polymeric shield. However, these frequencies are
related to the human voice. If teachers wore a face shield, the direct
sound would represent a minor contribution, compared to the
reflected sound. This may not be perceived as a comfortable situa- Mask ID 7 Mask ID 8
tion by the receivers, since they would be able to see the teacher in
a definite point in the indoor space but hear his/her voice coming
from other directions.

3.4. Overall considerations

Masks negatively affect voice propagation in classrooms,

regardless of the indoor acoustic conditions, but with differences
Mask ID 10
linked to characteristic male and female emissions. Furthermore, Mask ID 9
the results clearly identify how face shields split the emitted noise
into two different directions, thus compromising the direct
propagation.
The results above (Table 4) demonstrate that wearing masks
influences the male voice more than the female voice. Accordingly,
in the first case (male voice), more JNDs are found to be present
rather than in the second case (female voice) in both scenarios.
This depends on how the face masks behave when a sound wave
Fig. 12. Acoustic camera analyses in scenario A for all types of individual
propagates through them. A membrane-like model most appropri-
protections at 2000 Hz. Color legend represent 10 dB range. Center purple
ately explains the above results, with the mask stretched from nose represents the maximum level while external yellow-orange represents the
to mouth. In this case, a sound absorbing effect can be predicted minimum. (For interpretation of the references to color in this figure legend, the
when the resonance frequency of the system is verified, in accor- reader is referred to the web version of this article.)
dance with Eq. (1):
where f0 is the resonance frequency [Hz], d is the distance between
60
f 0 ¼ pffiffiffiffiffiffiffiffiffiffi ð1Þ the membrane and the solid behind it [m] and r is the superficial
dr mass of the membrane [kg/m2]. In this case, the membrane is the

9
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

Mask ID 1
STI male and female percentage A-weighted reducon
Mask ID 0
100
95

percentage A-weighted reducon [%]

90
85
80
75
70
65
60
55
Mask ID 2 Mask ID 3 50
125 250 500 1000 2000 4000
Frequency [Hz]

male female

Fig. 14. STI male and female percentage A-weighted trends, with reference to the
most impacting face protection frequency range (grey region).

may be determined based on real mask weights and average facial

differences, retrieved from the literature [64–66]. Thus, a range
Mask ID 4a Mask ID 4b
starting from about 1600 Hz to about 2200 Hz can be assessed. This
is clearly consistent with what is shown in Fig. 9, where the major
difference peak starts to rise at 1600 Hz, presents its maximum at
2000 Hz and then decreases. The noise limitation effect observed
at higher frequency ranges could be caused by sound absorption
of the textiles (masks) or transmission loss produced by transparent
polymer (face shields).
In this light, the morphology, shape and components of the face
protections act together to limit the sound wave generated by the
Mask ID 5 Mask ID 6
source. When analyzing the male–female voice difference, it may
be helpful to consider the A-weighted reference speech levels used
to calculate STI male and STI female. In Fig. 14, the STI male and
female -weighting trends (expressed in percentage of reduction)
are reported and combined with the range where the face protec-
tions mostly act.
It is evident how face protections act in a range where the
female voice presents a higher percentage reduction than the male
voice, with a difference of around 5% (3 dB in the standard). Thus,
Mask ID 7 Mask ID 8
the STI female is influenced less by wearing face protections than
the male STI, because the weighting penalizes the female voices
more, where the masks provide the most efficient sound reduction.

4. Conclusions

In this study, the influence of face protections on sound propa-

gation in classrooms was experimentally investigated. Two differ-
Mask ID 9 Mask ID 10 ent scenarios were considered, featuring very different indoor
sound fields (more reverberant, less reverberant). In both scenar-
ios, 10 different masks and face shield were tested. Speech intelli-
gibility index, sound pressure levels and emission lobes were
assessed.
The main findings are summarised as follows:

1) face protections worn by teachers in classrooms do affect

Fig. 13. Acoustic camera analyses in scenario B for all types of individual
protections at 2000 Hz. Center purple represents the maximum level while external
yellow-orange represents the minimum. (For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this article.)
mask, the solid is the mouth and the typical weight range is about
2–8 g. Masks are worn differently and this causes differences in the
membrane free area, however balanced by mass variations. In this
calculation, it is also difficult to define a perfect distance from the
mouth as well as the free vibrating areas related to single masks
due to the morphological diversity in human faces. However, ranges
sound propagation. Considering differences related to the
kind of protection device, speech transmission index could
vary up to 5 Just Noticeable Differences. In the sound
absorbing scenario (Scenario B), differences are higher than
in the reverberant one (Scenario A), both for the male voice
and all the receiver positions.
2) Faceprotectionsdefinedashighmedicalprotectionsaretheones
that mostlyreducesourceemission,exceptfor surgicalmasks.
3) Surgical masks, also featuring transparent plastic windows
on the mouth area for lip reading, are not shown to signifi-
cantly influence speech transmission in both classroom
scenarios.

10
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

4) Speech Transmission Index for male and female voices pre- CRediT authorship contribution statement
sent different reduction patterns. In particular, female
speech is less influenced by face protection compared to Marco Caniato: Conceptualization, Methodology, Measure-
male speech. This is due to the membrane resonance natural ments, Paper Writing, Supervision. Arianna Marzi: Measurements,
frequency and textile sound absorption. Paper Writing. Andrea Gasparella: Supervision.
5) The most affected frequencies lay in a range, which starts
form 1600 Hz and ends at 6300 Hz. The most affected octave
band is 2000 Hz for all face protections. In this frequency, a
Declaration of Competing Interest
minimum of 8 dB and a maximum of 14 dB reduction was
recorded, depending on the type of device worn and chosen
The authors declare that they have no known competing finan-
position in the classroom.
cial interests or personal relationships that could have appeared
6) All of the most influencing face protections present similar
to influence the work reported in this paper.
3D reduction patterns along the indoor acoustic field in both
scenarios, but with different values. It thus can be concluded
that the indoor sound field will be affected in the same way
in most of the cases (different face protection), but with Acknowledgements
some differences in noise reduction.
7) Source emission and directivity is affected by face protec- Authors would like to thank Christian Platzgummer for his pre-
tion, especially by face shields. In this case, mostly at high cious help (mask supports realization and panel movements).
frequencies, the emission lobe is always split into two parts Authors would also like to thank Chiara Da Rous for her kind
in both scenarios. This negatively affects sound perceptions, and wonderful work (mask realization and customization). This
as the source is not clearly identified in the space. work was supported by the Open Access Publishing Fund of the
Free University of Bozen-Bolzano.
Finally, it can be concluded that in classrooms surgical masks
should be adopted as they protect from mouth spray and do not
significantly reduce speech intelligibility and indoor sound field, Appendix A
both without and with a transparent window for lip reading. Other
types of tested facial masks significantly influence speech trans- In Tables A1–A6, in green the minimum STI value and in grey
missions and thus teacher-student communication in learning the maximum STI value for each position are depicted, in accor-
environments. dance with the methodology depicted in Section 3.1.

Table A1
Speech transmission index in reverberating and absorbing condition in position one.

ID Position ID Face Mask Scenario A Scenario B

STI Male STI Female STI Male STI Female
1 0 0.51 0.50 0.56 0.56
1 1 0.48 0.48 0.54 0.55
1 2 0.44 0.46 0.50 0.53
1 3 0.46 0.48 0.50 0.54
1 4a 0.46 0.47 0.48 0.52
1 4b 0.44 0.47 0.46 0.50
1 5 0.49 0.49 0.53 0.54
1 6 0.44 0.46 0.49 0.52
1 7 0.44 0.46 0.46 0.50
1 8 0.46 0.47 0.53 0.54
1 9 0.44 0.47 0.46 0.51
1 10 0.41 0.45 0.42 0.47

Table A2
Speech transmission index in reverberating and absorbing condition in position two.

ID Position ID Face Mask Scenario A Scenario B

STI Male STI Female STI Male STI Female
2 0 0.48 0.48 0.54 0.54
2 1 0.46 0.48 0.50 0.53
2 2 0.42 0.45 0.44 0.49
2 3 0.43 0.46 0.46 0.50
2 4a 0.41 0.44 0.46 0.51
2 4b 0.39 0.43 0.43 0.48
2 5 0.45 0.47 0.48 0.51
2 6 0.37 0.41 0.43 0.47
2 7 0.43 0.46 0.48 0.51
2 8 0.46 0.46 0.50 0.52
2 9 0.41 0.44 0.46 0.50
2 10 0.42 0.45 0.42 0.47

11
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

Table A3
Speech transmission index in reverberating and absorbing condition in position three.

ID Position ID Face Mask Scenario A Scenario B

STI Male STI Female STI Male STI Female
3 0 0.51 0.50 0.60 0.59
3 1 0.48 0.48 0.54 0.55
3 2 0.44 0.46 0.49 0.52
3 3 0.44 0.46 0.48 0.52
3 4a 0.47 0.46 0.51 0.53
3 4b 0.42 0.45 0.45 0.49
3 5 0.48 0.47 0.53 0.54
3 6 0.46 0.47 0.50 0.52
3 7 0.43 0.44 0.48 0.51
3 8 0.47 0.46 0.51 0.53
3 9 0.43 0.46 0.46 0.50
3 10 0.42 0.45 0.43 0.48

Table A4
Speech transmission index in reverberating and absorbing condition in position four.

ID Position ID Face Mask Scenario A Scenario B

STI Male STI Female STI Male STI Female
4 0 0.48 0.48 0.52 0.52
4 1 0.46 0.47 0.48 0.51
4 2 0.43 0.45 0.44 0.48
4 3 0.42 0.44 0.43 0.47
4 4a 0.43 0.45 0.47 0.50
4 4b 0.42 0.45 0.41 0.46
4 5 0.47 0.47 0.49 0.52
4 6 0.44 0.46 0.45 0.49
4 7 0.42 0.44 0.43 0.47
4 8 0.45 0.45 0.40 0.44
4 9 0.40 0.44 0.43 0.48
4 10 0.40 0.43 0.43 0.47

Table A5
Speech transmission index in reverberating and absorbing condition in position five.

ID Position ID Face Mask Scenario A Scenario B

STI Male STI Female STI Male STI Female
5 0 0.48 0.48 0.50 0.51
5 1 0.45 0.46 0.47 0.50
5 2 0.43 0.45 0.44 0.49
5 3 0.42 0.44 0.44 0.48
5 4a 0.44 0.46 0.44 0.48
5 4b 0.40 0.44 0.43 0.48
5 5 0.44 0.45 0.47 0.51
5 6 0.42 0.45 0.44 0.49
5 7 0.41 0.44 0.43 0.48
5 8 0.43 0.44 0.49 0.52
5 9 0.41 0.44 0.43 0.48
5 10 0.41 0.45 0.45 0.49

Table A6
Speech transmission index in reverberating and absorbing condition in position six.

ID Position ID Face Mask Scenario A Scenario B

STI Male STI Female STI Male STI Female
6 0 0.48 0.48 0.50 0.51
6 1 0.43 0.45 0.47 0.50
6 2 0.42 0.45 0.41 0.46
6 3 0.41 0.44 0.40 0.45
6 4a 0.40 0.43 0.43 0.48
6 4b 0.41 0.44 0.37 0.42
6 5 0.43 0.45 0.43 0.47
6 6 0.40 0.43 0.40 0.45
6 7 0.38 0.42 0.38 0.43
6 8 0.46 0.47 0.48 0.51
6 9 0.41 0.44 0.43 0.47
6 10 0.41 0.44 0.43 0.48

12
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

References [30] Skarlatos D, Manatakis M. Effects of classroom noise on students and teachers
in Greece. Percept Mot Skills 2003;96(2):539–44.
[31] Hodgson M, Rempel R, Kennedy S. Measurement and prediction of typical
[1] Coia JE, Ritchie L, Adisesh A, Makison Booth C, Bradley C, Bunyan D, Carson G,
speech and background-noise levels in university classrooms during lectures. J
Fry C, Hoffman P, Jenkins D, Healthcare Infection Society Working Group on
Acoust Soc Am 1999;105:226–33.
Respiratory and Facial Protection. Guidance on the use of respiratory and facial
[32] Crandell CC, Smaldino JJ, Flexer CA. Sound field amplification: applications to
protection equipment. J Hospital Infect 85 (n.d.); 2013: 170–182.
speech perception and classroom acoustics. Singular; 2005.
[2] Organization WH. Advice on the use of masks in the context of COVID-19:
[33] Bradley J. Classroom acoustics to support student learning. In: Jamieson D,
interim guidance. World Health Organization; 2020.
Rvachew S, Carhart R, Tillman TW, (1970). Interaction of competing speech
[3] Qaseem A, Etxeandia-Ikobaltzeta I, Yost J, Miller MC, Abraham GM, Obley AJ,
signals with hearing losses. Arch Otolaryn 91: 2007; 273–279.
et al. Use of N95, surgical, and cloth masks to prevent COVID-19 in health care
[34] Ronsse LM, Wang LM. Acoustics of elementary school classrooms: Correlations
and community settings: living practice points from the American college of
of varying background noise level and reverberation time to student
physicians (Version 1). Ann Intern Med 2020.
achievement. In: 38th International congress and exposition on noise control
[4] Fathizadeh H, Maroufi P, Momen-Heravi M, Dao S, Köse Sß , Ganbarov K, et al.
engineering 2009, INTER-NOISE 2009;2009. p. 1875–1879
Protection and disinfection policies against SARS-CoV-2 (COVID-19). Infez Med
[35] Shield B, Conetta R, Cox T, Mydlarz C, Dockrell J, Connolly D. Acoustics and
2020;28:185–91.
noise in English secondary schools. In: INTER-NOISE and NOISE-CON congress
[5] Brosseau LM, McCullough NV, Vesley D. Mycobacterial aerosol collection
and conference proceedings, Institute of Noise Control Engineering; 2013. p.
efficiency of respirator and surgical mask filters under varying conditions of
5672–5678.
flow and humidity. Appl Occup Environ Hyg 1997;12(6):435–45.
[36] Mikulski W, Radosz J. Acoustics of classrooms in primary schools–results of
[6] Marler H, Ditton A. ‘‘I’m smiling back at you”: exploring the impact of mask
the reverberation time and the speech transmission index assessments in
wearing on communication in healthcare. Int J Lang Commun Disorders
selected buildings. Arch Acoust 2011;36:777–93.
2021;56.
[37] Rościszewska T, Miśkiewicz A, Rogala T, Rudzki T, Fidecki T. Concert hall sound
[7] Carbon C-C. Wearing face masks strongly confuses counterparts in reading
clarity: a comparison of auditory judgments and objective measures. Arch
emotions. Front Psychol 2020;11:2526.
Acoust 2012:41–6.
[8] Ten Hulzen RD, Fabry DA. Impact of hearing loss and universal face masking in
[38] Cabrera D. Acoustic clarity and auditory room size perception. In: Cairns.
the COVID-19 era. In: Mayo clinic proceedings. Elsevier; 2020. p. 2069–72.
Australia 9-12 July, 2007; 2007.
[9] Ribeiro VV, Dassie-Leite AP, Pereira EC, Santos ADN, Martins P, de Alencar
[39] Zannin PHT, Zwirtes DPZ. Evaluation of the acoustic performance of
Irineu R. Effect of wearing a face mask on vocal self-perception during a
classrooms in public schools. Appl Acoust 2009;70(4):626–35.
pandemic. J Voice (2020).
[40] Astolfi A, Corrado V, Griginis A. Comparison between measured and calculated
[10] Eikenberry SE, Mancuso M, Iboi E, Phan T, Eikenberry K, Kuang Y, et al. To mask
parameters for the acoustical characterization of small classrooms. Appl
or not to mask: Modeling the potential for face mask use by the general public
Acoust 2008;69:966–76.
to curtail the COVID-19 pandemic. Infect Disease Model 2020.
[41] Prodi N, Visentin C, Feletti A. On the perception of speech in primary school
[11] Andrés JMA, de Castro MTGV, Vicente-Guijarro J, Peribáñez JB, Haro MG,
classrooms: Ranking of noise interference and of age influence. J Acoust Soc
Valencia-Martín JL, et al. Mascarillas como equipo de protección individual
Am 2013;133:255–68.
durante la pandemia de COVID-19: cómo, cuándo y cuáles deben utilizarse. J
[42] Seetha P, Karmegam K, Ismail MY, Sapuan SM, Ismail N, Moli L. Effects to
Healthcare Qual Res 2020;35:245–52.
teaching environment of noise level in school classrooms; 2008.
[12] Roy David. Masks method and impact in the classroom. Creative Educ 2020;11
[43] Radosz J. Global index of the acoustic quality of classrooms. Arch Acoust
(05):710–34.
2013;38:159–68.
[13] Wittum KJ, Feth L, Hoglund E. The effects of surgical masks on speech
[44] Bilzi P, Bozzoli F, Farina A. Influence of artificial mouth’s directivity in
perception in noise. In: Proceedings of meetings on acoustics ICA2013,
determining Speech Transmission Index, in: Audio Engineering Society
Acoustical Society of America; 2013. p. 060125.
Convention 119, Audio Engineering Society; 2005
[14] Goldin A, Weinstein B, Shiman N. How do medical masks degrade speech
[45] Stewart K, Cabrera D. Effect of acoustic environment on the sensitivity of
perception. Hear Rev 2020;27:8–9.
speech transmission index to source directivity. Arch Sci Rev 2009;52:71–8.
[15] Corey RM, Jones U, Singer AC. Acoustic effects of medical, cloth, and
[46] Tobey EA, Rekart D, Buckley K, Geers AE. Mode of communication and
transparent face masks on speech signals. ArXiv Preprint ArXiv:2008.04521.
classroom placement impact on speech intelligibility. Arch Otolaryngol-Head
(2020)
Neck Surg 2004;130:639.
[16] Palmiero AJ, Symons D, Morgan III JW, Shaffer RE. Speech intelligibility
[47] Van Engen KJ, Peelle JE. Listening effort and accented speech. Front Hum
assessment of protective facemasks and air-purifying respirators. J Occup
Neurosci 2014;8:577.
Environ Hygiene 2016;13(12):960–8.
[48] de Nobrega M, Opice R, Lauletta MM, de Nobrega CA. How face masks can
[17] Government of United Kingdom, Guidance Face coverings in education
affect school performance. Int J Pediatr Otorhinolaryngol 2020.
Updated 5 November 2020, Education and Childcare during Coronavirus (n.
[49] Atcherson SR, Finley ET, McDowell BR, Watson C. More Speech Degradations
d.). https://www.gov.uk/government/publications/face-coverings-in-
and Considerations in the Search for Transparent Face Coverings During the
education/face-coverings-in-education (accessed November 12, 2020).
COVID-19 Pandemic, Audiology Today. July/August 2020 (n.d.).
[18] The Victorian Government, Face masks in schools All school-based staff and
[50] I.S.O. Standard, 3382-1. Acoustics-measurement of room acoustic parameters-
students must wear face masks at school and when travelling to and from
part 1: performance spaces. International Organization for Standardization;
school., Face Masks in Schools (n.d.). https://www.coronavirus.vic.gov.au/face-
2009
masks-in-schools#use-of-personal-protective-equipment-in-education
[51] Tronchin L, Merli F, Dolci M. Virtual acoustic reconstruction of the Miners’
(accessed November 12, 2020).
Theatre in Idrija (Slovenia). Appl Acoust 2021;172:107595. https://doi.org/
[19] U.S. Department of Health & Human Services, Cloth face masks in schools,
10.1016/j.apacoust.2020.107595.
Public Healt Emergency (2020). https://www.phe.gov/facecovering/
[52] Tronchin L, Merli F, Manfren M. On the acoustics of the Teatro 1763 in Bologna.
Pages/cloth-face-masks-in-school.aspx (accessed November 12, 2020).
Appl Acoust 2021;172:107598. https://doi.org/10.1016/j.
[20] Spitzer M. Masked education? The benefits and burdens of wearing face masks
apacoust.2020.107598.
in schools during the current Corona pandemic. Trends Neurosci Educ 20:
[53] IEC 60268: Sound system equipment–Part 16: Objective rating of speech
2020; 100138.
intelligibility by speech transmission index, (2003).
[21] GBS Buildings, D. Unit, Acoustic design of schools, Stationery Office; 2003.
[54] P. CODE, Sound system equipment–Part 16: Objective rating of speech
[22] Klatte M, Lachmann T, Meis M. Effects of noise and reverberation on speech
intelligibility by speech transmission index; 2003.
perception and listening comprehension of children and adults in a classroom-
[55] Steeneken HJM, Houtgast T. A physical method for measuring speech-
like setting. Noise Health 2010;12:270.
transmission quality. J Acoust Soc Am 1980;67:318–26.
[23] Shield B, Conetta R, Dockrell J, Connolly D, Cox T, Mydlarz C. A survey of
[56] Mapp P. Uncertainties in speech transmission index measurements. J Acoust
acoustic conditions and noise levels in secondary school classrooms in
Soc Am 2014;135. 2236–2236.
England. J Acoust Soc Am 2015;137(1):177–88.
[57] Word Health organization, Mask use in the context of COVID-19: Interim
[24] D. for E. and S. United Kingdom, Building Bulletin 93 : Acoustic Design of
guidance, Word Health Organization. (2020). https://www.who.int/
Schools., DfES London; 2003
publications/i/item/advice-on-the-use-of-masks-in-the-community-during-
[25] Lubman David, Sutherland Louis C. Good classroom acoustics are a good
home-care-and-in-healthcare-settings-in-the-context-of-the-novel-
investment for America. J Acoust Soc Am 1999;106(4). 2283–2283.
coronavirus-(2019-ncov)-outbreak (accessed December 1, 2020).
[26] Pekkarinen Eeva, Viljancn Vesa. Acoustic conditions for speech communication
[58] W. World Health Organization, Coronavirus disease (COVID-19): Schools,
in classrooms. Scand Audiol 1991;20(4):257–63.
World Health Organization. (2020). https://www.who.int/emergencies/
[27] Ljung R, Sörqvist P, Kjellberg A, Green A-M. Poor listening conditions impair
diseases/novel-coronavirus-2019/question-and-answers-hub/q-a-detail/q-a-
memory for intelligible lectures: implications for acoustic classroom
schools-and-covid-19?gclid=CjwKCAjw8-
standards. Build Acoust 2009;16:257–65.
78BRA0EiwAFUw8LB_KDmwWvvY7HKs42SwpEJnSRVige8tiwLix2-
[28] Garnier M, Henrich N. Speaking in noise: How does the Lombard effect
MNwPbrZ3UJ_sGsKRoC7GgQAvD_BwE.
improve acoustic contrasts between speech and ambient noise? Comput
[59] Seraphim H-P. Untersuchungen über die Unterschiedsschwelle exponentiellen
Speech Lang 2014;28:580–97.
Abklingens von Rauschbandimpulsen. Acta Acust United Acust 1958;8:280–4.
[29] Sarampalis A, Kalluri S, Edwards B, Hafter E. Objective measures of listening
[60] Karjalainen M, Jarvelainen H. More about this reverberation science:
effort: Effects of background noise and noise reduction. J Speech Lang Hear Res
perceptually good late reverberation. In: Audio engineering society
2009.
convention 111, Audio Engineering Society; 2001

13
M. Caniato, A. Marzi and A. Gasparella Applied Acoustics 178 (2021) 108051

[61] Bradley JS, Reich R, Norcross SG. A just noticeable difference in C50 for speech.
Appl Acoust 1999;58(2):99–108.
[62] Mealings K. Classroom acoustic conditions: Understanding what is suitable
through a review of national and international standards, recommendations,
and live classroom measurements; 2016.
[63] Barnett PW, Acoustics AM. Overview of speech intelligibility. Proc Inst Acoust
1999;21:1–16.
[64] Heidari Z, Sagheb H-RM. Morphological evaluation of head and face in 18–25
years old. J Med Sci 2006;6:462–6.
[65] Abbas H. 3D face morphology classification for medical applications, Cardiff
University; 2018.
[66] Tronchin L. Variability of room acoustic parameters with thermo-hygrometric
conditions. Appl Acoust 177: 2021; 107933

14