Content uploaded by Liran Oren
Author content
All content in this area was uploaded by Liran Oren on Jan 03, 2024
Content may be subject to copyright.
TaggedH1How Face Masks Affect Acoustic and Auditory Perceptual
Characteristics of the Singing VoiceTaggedEnd
TaggedP*Liran Oren,
†
Michael Rollins,
‡
Ephraim Gutmark, and *Rebecca Howell, *yzCincinnati, OhioTaggedEnd
TaggedPSummary: Wearing a face mask has been accepted as one of the most effective ways for slowing the spread of
COVID-19. Yet information regarding the degree to which masks affect acoustics and perception associated with
voice performers is scarce. This study examines these effects with common face masks, namely a neck gaiter, dis-
posable surgical mask, and N95 mask, as well as a novel material that could be used as a mask (acoustic foam).
A recorded excerpt from the “Star-Spangled Banner”was played through a miniature speaker placed inside the
mouth of a masked manikin. Experienced listeners were asked to rate perceptual qualities of these singing stimuli
by blindly comparing them with the same recording captured without a mask. Acoustic analysis showed that face
masks affected the sound by enhancing or suppressing different frequency bands compared to no mask. Acoustic
energy around the singer’s formant was reduced when using surgical and N95 masks, which matches observations
that these masks are more detrimental to the perceptions of singing voice compared with neck gaiter or acoustic
foam. It suggests that singers can benefit from masks designed for minimal impact on auditory perception of the
singing voice while maintaining reasonable efficacy of filtering efficiency.TaggedEnd
TaggedPKey Words: Singing voice−Face mask−Acoustics−Perceptual characteristics.TaggedEnd
TAGGEDH1INTRODUCTIONTAGGEDEND
TaggedPBroadway theatre in New York City is a billion-dollar
industry that employs thousands and contributes billions
more to the local economy in tourism.
1
On March 17, 2020
the Skagit County, Washington choir informed public
health officials of ill members, leading to 53 of 122 testing
positive for SARS-CoV-2 virus. Since this report, the act of
singing has been identified as a contributor of transmission
(emission of aerosols by loudness of vocalization) or super-
emitters (those that release more aerosols compared to
others).
2
TaggedEnd
TaggedPThe mode of transmission for SARS-CoV-2 virus is expo-
sure to infected respiratory droplets or aerosolized par-
ticles.
3
Both types of particles can be expelled by infected
individuals through coughing, sneezing, speech, singing, or
breathing.
4
Droplets are particles that, once emitted by a
person, will follow a trajectory that is affected predomi-
nantly by gravity and depends on their size and ejection
force. Aerosolized particles are generally <5mm and can
remain suspended in the air for longer periods after being
expelled.
4,5
The transmission of SARS-CoV-2 occurs when
healthy individuals come into contact with or inhale par-
ticles that carry these pathogens.
6
TaggedEnd
TaggedPUntil herd immunity can be achieved (likely by vaccina-
tion), public health efforts to combat COVID-19 have
focused on prevention strategies. The Center for Disease
Control has focused on reducing transmission via droplet
and aerosolized particles by encouraging social distancing,
avoiding crowds, hand washing, and mask wearing. Wear-
ing a facial covering has become a common practice to miti-
gate transmission.TaggedEnd
TaggedPAll face masks, whether home-made or well-fitted (eg,
medical grade), can effectively block the larger droplet size
particles. However, the filtration efficiency, defined as the
percentage of aerosolized particles stopped by the mask
from spreading away from the source, varies with the filter
material. For example, the filter efficiency of woven or non-
woven polypropylene, such as a disposable surgical mask or
N95 mask, is about 50% and 95%, respectively, compared
with about 10% efficiency for fabric material such cotton.
7
−9
Despite these differences, it is well documented that using
any type of face covering is critical to reducing the number
of hospitalizations and deaths related to COVID-19.
10−12
TaggedEnd
TaggedPThe factors affecting the filtration efficiency of masks are
well documented, but their impact on speech and especially
singing has not been sufficiently quantified. Studies have
shown that masks can reduce both intelligibility and loud-
ness of speech
13−15
and that speakers wearing a facial cover-
ing tend to make their normal speech louder.
16
Similar
information regarding the degree to which face masks affect
acoustics and perception associated with professional voice
users such as singers is scarce. With live entertainment
largely on hold, wearing face masks will likely be recom-
mended once these activities resume in efforts to minimize
risks of transmitting COVID-19, especially considering it is
known that loud speech increases emission of aerosolized
particles.
17,18
Because vocal performers rely on the percep-
tual qualities of their voice
19
(eg, loudness, clarity, and
ring), there is a need to improve the understanding regard-
ing the acoustic effects of masks and auditory perception of
the singing voice.TaggedEnd
TaggedPAs a first step, this study examines the effects of common
face masks, namely a neck gaiter, disposable surgical mask,
TaggedEndAccepted for publication February 22, 2021.
TaggedEndTaggedEndThe authors have no financial or personal interests that could inappropriately influ-
ence (bias) the work in this manuscript.
TaggedEndFrom the *Department of Otolaryngology-Head and Neck Surgery, University of
Cincinnati, Cincinnati, Ohio; TaggedEndyDepartment of Biomedical Engineering, University of
Cincinnati, Cincinnati, Ohio; and the TaggedEndzDepartment of Aerospace Engineering and
Engineering Mechanics, University of Cincinnati, Cincinnati, Ohio.
TaggedEndAddress correspondence and reprint requests to Liran Oren, Department of Otolar-
yngology-Head and Neck Surgery, University of Cincinnati, 231 Albert Sabin Way
Room 6113, MSB PO Box 670528, Cincinnati, OH 45267. E-mail: liran.oren@uc.edu
TaggedEndJournal of Voice, Vol. 37, No. 4, pp. 515−521
TaggedEnd0892-1997
TaggedEnd© 2021 The Voice Foundation. Published by Elsevier Inc. All rights reserved.
TaggedEndhttps://doi.org/10.1016/j.jvoice.2021.02.028
and N95 mask, as well as a novel material that can be used
as a mask (acoustic foam), on acoustics measures and per-
ceptual characteristics of listening to a masked singing
voice. A recorded singing voice was used as the acoustic
source in order to isolate the direct acoustic effects caused
by the mask from the influences of using a mask on the act
of singing. This study therefore does not evaluate the sing-
er’s perspective, only the listener’s.TaggedEnd
TAGGEDH1METHODSTAGGEDEND
TaggedH2Setup and recording of singing voice stimuliTaggedEnd
TaggedPThe audio sample used for the study consisted of the phrase
“O'er the land of the free”extracted from a recording of a
soprano singing “The Star-Spangled Banner”a cappella. It
was played from a miniature speaker (Intsun, Tomtop Elec-
tronic) that was placed inside the mouth of an airway simu-
lator manikin (AirSim Combo Bronchi X, TruCorp). The
manikin’s external features such as head circumference,
nose and ears matched an average human adult. The minia-
ture speaker was aligned with the manikin’s lips and was
connected to a laptop by an audio cable that was threaded
through the airway (Figure 1a). This setup intended to
mimic a reproducible singing voice emanating from a singer
and also allowed for comparable evaluation of several types
of masks.TaggedEnd
TaggedPRecordings of the audio sample were captured with the
manikin’s face covered with one of three commonly used
face masks (Figures 1b-d). These include a neck gaiter (sin-
gle-layer cotton fabric material), a disposable surgical mask,
and an N95 mask. Reference data were obtained by record-
ing a sample without the mask. In addition to the three
standard masks and the reference, the manikin’sfacewas
covered with acoustic foam of a single- or double-layer (7.5
and 15 mm, respectively) thickness (Figure 1e). Overall, the
audio recordings were made for six mask configurations.TaggedEnd
TaggedH2Audio recordingsTaggedEnd
TaggedPThe manikin was placed near the center of an anechoic
chamber (7.6 by 7.2 m). A 1/2-inch microphone (Type 2671,
Br€
uel & Kjær) was positioned 30 cm directly in front of the
mouth. In testing directivity effects of this setup, there was a
<3 dB change when the microphone was offset by 30° from
the sagittal plane. All audio recordings were captured at
51.2 kHz using a National Instruments data acquisition sys-
tem (NI 9234).TaggedEnd
TaggedPSinging audio samples were captured for 6.6 seconds with
each mask. Recordings with each mask type also included a
chirp signal with a tone that increased logarithmically over
the frequency range from 20 to 20 kHz over 1.5 seconds.
The chirp sound was repeated three times (total of 4.5 sec-
onds) in each recording.TaggedEnd
TaggedH2Acoustical data analysisTaggedEnd
TaggedPThe acoustic analysis was based on the spectrum for the
audio data captured with the chirp signal. The analysis also
included extracting 0.5 seconds of the sustained /æ/ and /i/
vowel segments in each singing sample and computing the
acoustic energy of the singer’s formant
20
by integrating the
TaggedEnd TaggedFigure
FIGURE 1. Setup used for audio recordings. Miniature speaker is placed inside the manikin’s mouth and the same audio sample is played
with five types of masks. (a) Baseline recording with no face mask, (b) neck gaiter, (c) disposable surgical mask, (d) N95 mask, (e) acoustic
foam. Single- (shown) and double-layers of acoustic foam were tested.TaggedEnd
TaggedEnd TaggedFigure
FIGURE 2. Spectra of /æ/ vowel segment extracted from the
recorded samples of baseline (no mask) and N95 mask. Acoustic
energy in the singer’s formant energy is computed by integrating
the area under the curve between 2 and 3.5 kHz (shown for the no-
mask case as the shaded area).TaggedEnd
TaggedEnd516 Journal of Voice, Vol. 37, No. 4, 2023
spectral energy in the frequency band of 2−3.5 kHz
(Figure 2).TaggedEnd
TaggedH2Perceptual evaluationsTaggedEnd
TaggedPA panel of four listeners (R1-R4) rated the recorded singing
samples for each of the six mask configurations. All four
have personal backgrounds in vocal performance and
speech-language pathology training (range of 0−7 years
clinical experience). Each listener completed the evaluation
alone seated in a quiet office in front of a laptop that pre-
sented the audio samples through its internal loudspeakers.TaggedEnd
TaggedPPerceptual ratings were done using a paired-stimulus pre-
sentation method. That is, each listener rated a series of
trials, each trial containing a pair of singing stimuli. The
presentation of audio recordings and capture of the percep-
tual response were done using the Alvin program.
21
Listen-
ers could control the pace of presentation and their response
time for each trial. Although they were allowed to repeat
each pair as often as needed, they had to make their selec-
tion before advancing to the next pair. The samples in each
pair were separated by 0.4 second pause.TaggedEnd
TaggedPThe paired-stimulus had listeners compare a baseline
recording (no mask) with another recording in a blinded
fashion. Rating the singing stimulus of each mask was
repeated 8 times, with the baseline recording presented
4 times as the first of the pair and 4 times as the second
sample. The paired samples also included comparison of the
baseline with itself; this configuration was done to aid in
assessing rater reliability and only repeated 4 times. In sum-
mary, each listener rated a total of 44 singing samples (five
masks for 8 times + no mask for 4 times). The order of pre-
sentation was randomized and the experiment lasted 10
−15 minutes for each listener.TaggedEnd
TaggedPThe perceptual evaluations included comparison of the
“ring”and “overall singing quality”for each pair. The com-
puter displayed a “slider scale,”with the slider button posi-
tioned at the midpoint (Figure 3). The ends of the slider
scale were labeled as Sample 1 and Sample 2, respectively.
The following instructions were read to listeners before the
experiment: “You will be presented with a series of paired
audio segments sung by a woman wearing a face mask. For
each presentation, the segments should have differing sing-
ing qualities. Your task is to (1) indicate whether Sample 1
TaggedEnd TaggedFigure
FIGURE 3. Screen capture of the Alvin program used for perceptual ratings. Baseline recording was always included as one of the pair.
Rating was done by having the listener moving the slider toward the side with better “Ring”and “Overall Singing Quality.”TaggedEnd
TaggedEndLiran Oren, et al How Face Masks Affect Acoustic and Auditory Perceptual Characteristics of the Si TaggedEnd517
or Sample 2 has the better Ring and (2) by how much.
Use the mouse to move the slider from the middle of the
scale toward the presentation that has the better Ring. Indi-
cate the degree of Ring difference by how close you move
the marker to the speech sample; the closer to either end,
the larger the difference in Ring. You will then perform
a similar judgment of the Overall Singing Quality on the
second slider.”TaggedEnd
TaggedPThe Alvin program recorded the listener’s rating for each
pair as an integer that ranged from 500 to +500, where
“minus”scores indicate better “ring”or “overall singing
quality”of Sample 1 while “plus”scores related to Sample
2. For data analysis, the score’s absolute value was used as
it was always referenced to the “no mask”baseline. A rating
of zero would indicate that there was no difference between
the two samples. Hence, lower ratings indicated that the
singing recording with a mask was perceived to be more like
the singing without a mask (i.e., lower score means less
impact of the mask).TaggedEnd
TaggedPSample pairs where listeners judged the baseline sample
(no mask) to have worse ring or worse overall singing qual-
ity than the singing sample with a mask were removed from
the data set (27 of the total 176 samples, 15%). The statisti-
cal mean (and its 95% confidence interval) of the ratings of
the singing samples with each mask of the 4 listeners was
then computed. Rater reliability was assessed using the
intraclass correlation coefficient, ICC(2,k).
22
TaggedEnd
TAGGEDH1RESULTSTAGGEDEND
TaggedPAcoustic analysis showed that the face masks affected the
sound by enhancing or suppressing different frequency
bands compared to no mask. For example, the difference
between N95 mask and the baseline spectra are shown in
Figure 4a. The difference between each mask and baseline
was assessed by subtracting the spectrum of the chirp signal
for baseline from the spectrum of each mask (Figure 4b).
Because the chirp signal includes all audible frequencies
(within the 20 Hz to 20 kHz range), the differences between
the spectra showed which frequencies were amplified and
which were suppressed for each mask. Compared with base-
line findings, the neck gaiter had a small effect on the spec-
trum with some amplification (<2.5 dB) of frequencies
below 4 kHz and minimal differences for frequencies
>4 kHz. Amplification of frequencies below 2 kHz was
observed for all masks. The greatest amplification (~5 dB)
was observed with the disposable surgical mask and with
two layers of acoustic foam. The surgical mask also showed
significant amplification also between 4.5 and 7.0 kHz.TaggedEnd
TaggedPThe face masks caused suppression in other frequency
bands. Most notably, the N95 mask suppressed frequencies
between 2 and 5 kHz and above 6 kHz. The surgical mask
showed some suppression between 3-5 kHz. The neck gaiter
yielded minimal change in the spectrum. Similar spectral
trends for each mask were also found using the long time
averaging spectrum of the singing audio samples.TaggedEnd
TaggedPPerceptual testing shows a direct relationship between the
type of mask and the perception of the singing samples
(Figure 5). Overall, the perception of “ring”seemed to be
more affected when using a mask than the perception of
‘overall singing quality’. Three out of four listeners showed
excellent intra-rater reliability (0.98 for R2, 0.93 for R3,
and 0.91 for R4). Intrarater reliability for R1 was 0.71,
which is moderate, perhaps because of lack of experience in
performing perceptual ratings. The reliability between lis-
teners was moderate at 0.67, which was similar to the
TaggedEnd TaggedFigure
(a) (b)
FIGURE 4. Face masks amplify and supressed different frequencies. (a) Spectra of the audio sample using chirp singal; only the baseline
and N95 cases are shown. (b) Difference of each mask’s spectra relative to baseline (no mask) using a chirp signal.TaggedEnd
TaggedEnd518 Journal of Voice, Vol. 37, No. 4, 2023
current literature using perceptual ratings of the singing
voice.
23-25
TaggedEnd
TaggedPThe statistical mean (with its 95% confidence interval) for
the listener’s ratings of each mask is shown in Figure 6 (left
ordinate). The figure also shows the corresponding acoustic
energy calculated for the 2−3.5 kHz frequency band (the
singer’s formant) with each mask (right ordinate). Overall,
there was minimal difference between no mask and a neck
gaiter while a disposable surgical mask or an N95 mask was
more detrimental to ring and overall singing quality percep-
tions. The N95 mask had the greatest impact on reducing
the ring quality. Single layer of acoustic foam had minimum
impact on perception compared to both surgical and N95
masks.TaggedEnd
TAGGEDH1DISCUSSIONTAGGEDEND
TaggedPThis study provides useful information regarding the impact
of different face masks on the acoustics and perceptual
measures of the singing voice for the listener. This will be of
particular interest to voice professionals such as singers,
actors, and vocal coaches. A mask creates a barrier to sound
propagation, acting as an acoustic filter where material and
thickness determine its response characteristics. Changes in
TaggedEnd TaggedFigure
(a) (b)
FIGURE 5. Perceptual ratings of singing samples by mask type. Lower scores indicate less perceptual difference relative to baseline (no
mask). R1−R4 correspond to each listener (a) Ring. (b) Overall singing quality.TaggedEnd
TaggedEnd TaggedFigure
(a) (b)
FIGURE 6. Comparing perceptual and acoustic characterisitcs of all mask configurations. The statiscal mean (with 95% confidance inter-
val) based on all listeners’ratings for each mask is shown on the left ordinate for the perception of (a) “Ring”and (b) “Overall singing qual-
ity.”The acoustic energy within the 2−3.5 kHz frequency band (the singer’s formant) during sustained /æ/ and /i/ vowel segments in each
singing sample is shown on the right ordinate of each plot.TaggedEnd
TaggedEndLiran Oren, et al How Face Masks Affect Acoustic and Auditory Perceptual Characteristics of the Si TaggedEnd519
these factors can amplify or attenuate certain frequency
bands, thus altering the formants and affecting the percep-
tion of radiated sound.TaggedEnd
TaggedPThe purpose of the face mask is to maintain a reasonable
efficacy in filtering aerosolized particles. An ideal singing
mask should be designed to avoid frequency suppression
around the signer’s formant and be made from material(s)
with the highest filter efficiency.TaggedEnd
TaggedPA neck gaiter has minimal effect compared to no face
mask. However, a single-layer cotton material common to
neck gaiter design offers minimal filtering efficiency, specifi-
cally for aerosolized particles. While the N95 mask is highly
effective in filtering air particles, this study suggests it is far
more disruptive to the auditory perceptual characteristics of
the singing voice, possibly because it suppresses the singer’s
formant.TaggedEnd
TaggedPThe magnitude of acoustic energy in the singer’s formant
frequency band could be an indicator of the perceptual data
trends. There was minimal difference in acoustic energy
between the no-mask baseline, neck gaiter, and acoustic
foams. The bigger differences for the surgical and N95
masks matched the drop in their perceptual ratings.
Changes to the acoustic energy in the singer’s formant are
expected by looking at the chirp signal spectra. The differ-
ence from baseline (Figure 4b) shows both surgical and N95
masks suppressed some of the frequencies within the singer’s
formant, with the latter to a larger extent. This comparison
shows that the amplification of frequencies below 2 kHz,
which occurred with all masks, did not affect the perception.
Although the highest amplification occurred for the double
layer foam, the perceptual difference relative to the baseline
was not as pronounced as for the N95 mask.TaggedEnd
TaggedPAcoustic foam, which is not a material currently used for
face covering, showed little effect on the perception of sing-
ing voice. Its single- and double-layer filtration efficiency is
about 30% and 50%, respectively.
26
These values are similar
to disposable surgical masks but far lower than N95. Using
acoustic foam as face covering material would require fur-
ther characterization such as its breathing resistance
27
(ie,
difficulty of breathing through the material). An ideal sing-
ing mask should be designed to avoid frequency suppression
around the signer’s formant and be made from material(s)
with the highest filter efficiency.TaggedEnd
TaggedH2LimitationsTaggedEnd
TaggedPThere are several limitations to our study. First, we only
considered a limited number of masks and only certain
types. For example, neck gaiter with multiple layers, thicker
N95 material, and surgical mask with medical grade can
absorb the sound differently and therefore will have differ-
ent perception and/or acoustic characteristics. Second, the
sample recording was a soprano voice of a native-English
female singer. A singer’s formant is known to depend on the
singing range (alto/tenor/baritone/bass).
28
Third, the acous-
tic propagation from the computer’s internal speakers can
limit listeners’ability to notice perceptual differences. This
could also explain why nearly 15% of the ratings were “mis-
judged”(as if wearing these specific masks is better com-
pared with no mask) and needed to be removed from the
analysis. It is reasonable to expect different perceptual
results if the listeners were using high-quality headphones.
Lastly, the study did not consider perception from the singer
using a mask; this would require a different analysis on the
aerodynamics and compensatory breathing patterns of a
singer wearing a mask. However, it is possible that profes-
sional singers could adapt their singing to the mask, allow-
ing their voice to be perceived as the same quality by a
listener, regardless of the mask type used. This requires fur-
ther evaluation.TaggedEnd
TaggedH1ACKNOWLEDGMENTSTaggedEnd
TaggedPThe authors would like to thank Sergey Grinshpun, PhD for
measuring the filter efficiency of the acoustic foam material.TaggedEnd
TAGGEDH1REFERENCESTAGGEDEND
TaggedP 1. Grosses - Broadway in NYC. The Broadway League. www.broadway
league.com/research. Accessed December 09, 2020.TaggedEnd
TaggedP 2. Hamner L, Dubbel P, Capron I, et al. High SARS-CoV-2 Attack Rate
Following Exposure at a Choir Practice —Skagit County, Washing-
ton, March 2020. Morbid Mort Wkl Rep (MMWR). 2020;69:606–
610. https://doi.org/10.15585/mmwr.mm6919e6.TaggedEnd
TaggedP 3. Scientific Brief: SARS-CoV-2 and Potential Airborne Transmission.
Coronavirus Disease 2019 (COVID-19). Vol 2020: Centers for Disease
Control.TaggedEnd
TaggedP 4. Morawska L, Johnson G, Ristovski Z, et al. Size distribution and sites
of origin of droplets expelled from the human respiratory tract during
expiratory activities. J Aerosol Science. 2009;40(3):256–269. https://
doi.org/10.1016/j.jaerosci.2008.11.002.TaggedEnd
TaggedP 5. Duguid JP. The size and the duration of air-carriage of respiratory
droplets and droplet-nuclei. Epidemiol Infect. 1946;44(6):471–479.
https://doi.org/10.1017/S0022172400019288.TaggedEnd
TaggedP 6. Van Doremalen N, Bushmaker T, Morris DH, et al. Aerosol and sur-
face stability of SARS-CoV-2 as compared with SARS-CoV-1. N Engl
J Med. 2020;382:1564–1567. https://doi.org/10.1056/NEJMc2004973.TaggedEnd
TaggedP 7. Pan J, Harb C, Leng W, et al. Inward and outward effectiveness of
cloth masks, a surgical mask, and a face shield. Aerosol Sci Tech.
2021:1–16. https://doi.org/10.1080/02786826.2021.1890687.TaggedEnd
TaggedP 8. Tcharkhtchi A, Abbasnezhad N, Seydani MZ, et al. An overview of fil-
tration efficiency through the masks: mechanisms of the aerosols pene-
tration. Bioactive Mat. 2021;6(1):106–122. https://doi.org/10.1016/j.
bioactmat.2020.08.002.TaggedEnd
TaggedP 9. Asadi S, Cappa CD, Barreda S, et al. Efficacy of masks and face cover-
ings in controlling outward aerosol particle emission from expiratory
activities. Sci Rep. 2020;10(1):1–13. https://doi.org/10.1038/s41598-
020-72798-7. 15665.TaggedEnd
TaggedP10. Eikenberry SE, Mancuso M, Iboi E, et al. To mask or not to mask:
modeling the potential for face mask use by the general public to cur-
tail the COVID-19 pandemic. Infect Dis Model. 2020;5:293–308.
https://doi.org/10.1016/j.idm.2020.04.001.TaggedEnd
TaggedP11. Liang M, Gao L, Cheng C, et al. Efficacy of face mask in preventing
respiratory virus transmission: a systematic review and meta-analysis.
Travel Med Infect Dis. 2020;36: 101751. https://doi.org/10.1016/j.
tmaid.2020.101751.TaggedEnd
TaggedP12. Gandhi M, Beyrer C, Goosby E. Masks do more than protect others
during COVID-19: reducing the inoculum of SARS-CoV-2 to protect
the wearer. J Gen Intern Med. 2020;35:3063–3066. https://doi.org/
10.1007/s11606-020-06067-8.TaggedEnd
TaggedP13. Palmiero AJ, Symons D, Morgan III JW, et al. Speech intelligibility
assessment of protective facemasks and air-purifying respirators. J
TaggedEnd520 Journal of Voice, Vol. 37, No. 4, 2023
Occup Environ Hyg. 2016;13:960–968. https://doi.org/10.1080/
15459624.2016.1200723.TaggedEnd
TaggedP14. Mendel LL, Gardino JA, Atcherson SR. Speech understanding using
surgical masks: a problem in health care? J Am Acad Audiol. 2008;19
(9):686–695. https://doi.org/10.3766/jaaa.19.9.4.TaggedEnd
TaggedP15. Saeidi R, Huhtakallio I, Alku P. Analysis of face mask effect on
speaker recognition. Interspeech. 2016:1800–1804. 2016. https://www.
isca-speech.org/archive/Interspeech_2016/pdfs/0518.PDFTaggedEnd. Accessed
March 2021.
TaggedP16. Fecher N. Effects of Forensically-Relevant Facial Concealment on
Acoustic and Perceptual Properties of Consonants. University of York;
2014. Doctoral dissertation; 2014. https://etheses.whiterose.ac.uk/
7397/1/Natalie_Fecher_PhD_2014.pdfTaggedEnd. Accessed March 2021.
TaggedP17. Asadi S, Wexler AS, Cappa CD, et al. Effect of voicing and articulation
manner on aerosol particle emission during human speech. PLoS One.
2020;15:(1) e0227699. https://doi.org/10.1371/journal.pone.0227699.TaggedEnd
TaggedP18. Asadi S, Wexler AS, Cappa CD, et al. Aerosol emission and super-
emission during human speech increase with voice loudness. Sci Rep.
2019;9:1–10. https://doi.org/10.1038/s41598-019-38808-z. 2348.TaggedEnd
TaggedP19. Ekholm E, Papagiannis GC, Chagnon FP. Relating objective measure-
ments to expert evaluation of voice quality in western classical singing:
critical perceptual parameters. J Voice. 1998;12(2):182–196. https://
doi.org/10.1016/S0892-1997(98)80038-6.TaggedEnd
TaggedP20. Sundberg J, Rossing TD. The Science of Singing Voice. Northern Illi-
nois University Press; 1987.TaggedEnd
TaggedP21. Hillenbrand JM, Gayvert RT. Open source software for experiment
design and control. J Speech Lang Hear Res. 2005;48(1):45–60. https://
doi.org/10.1044/1092-4388(2005/005).TaggedEnd
TaggedP22. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater
reliability. Psychol Bull. 1979;86(2):420–428. https://doi.org/10.1037/
0033-2909.86.2.420.TaggedEnd
T
aggedP23. Eckert DJ, White DP, Jordan AS, et al. Defining phenotypic causes of
obstructive sleep apnea. Identification of novel therapeutic targets. Am
J Respir Crit Care Med. 2013;188:996–1004. https://doi.org/10.1164/
rccm.201303-0448OC.TaggedEnd
TaggedP24. Gupta C, Li H, Wang Y. Perceptual evaluation of singing quality.
2017 Asia-Pacific Signal and Information Processing Association
Annual Summit and Conference (APSIPA ASC). IEEE; 2017:577–
586. https://doi.org/10.1109/APSIPA.2017.8282110.TaggedEnd
TaggedP25. Oates JM, Bain B, Davis P, et al. Development of an auditory-percep-
tual rating instrument for the operatic singing voice. J Voice. 2006;20
(1):71–81. https://doi.org/10.1016/j.jvoice.2005.01.006.TaggedEnd
TaggedP26. Filtration efficiency of acoustic foam materials. Unpublished results.TaggedEnd
TaggedP27. O’Kelly E, Pirog S, Ward J, et al. Ability of fabric face mask materials
to filter ultrafine particles at coughing velocity. BMJ Open. 2020;10:(9)
e039424. https://doi.org/10.1136/bmjopen-2020-039424.TaggedEnd
TaggedP28. Sundberg J. Level and center frequency of the singer’s formant.
J Voice. 2001;15(2):176–186. https://doi.org/10.1016/S0892-1997(01)
00019-4.TaggedEnd
TaggedEndLiran Oren, et al How Face Masks Affect Acoustic and Auditory Perceptual Characteristics of the Si TaggedEnd521
