Acoustical Effects of Surgical and N95 Masks on Speech Perception in

Open-plan Offices Pooja Ganatra Buro Happold 3001 Rolex Tower, DIFC, Dubai

ABSTRACT

Wearing face masks (alongside physical distancing) provides some protection against COVID-19. Face masks can also change how people communicate and subsequently affect speech signal qual- ity. This study investigated how two common face mask types affect acoustic analysis of speech per- ception. Quantitative and qualitative assessments were carried out in terms of measuring the sound pressure levels and playing back to a group of people. The responses gauged proved that masks al- ter the speech signal with downstream effects on speech intelligibility of a speaker. Masks muffle speech sounds at higher frequencies and hence the acoustic effect of a speaker wearing a face mask is equivalent to the listener having slight high frequency hearing loss. When asked on the percep- tion of audibility, over 83% of the participants were able to clearly hear the no mask audio clip, however, 41% of the participants thought it was moderately audible with N95 and face shield masks. Due to no visual access, face masks act as communication barriers with 50% of the people finding to understand people because they could not read their lips. Nevertheless, based on these findings it is reasonable to hypothesize that wearing a mask would attenuate speech spectra at simi- lar frequency bands. 1. INTRODUCTION

As the world works to control the novel coronavirus 2019 (COVID-19) pandemic, face masks are expected to prove critical to slowing the spread of the virus. Extensive research proposes that indi- viduals who wear a face cover fundamentally bring down the danger of spreading COVID-19 to others through talking. Talking without face security opens individuals close by to infection laden drops that would somehow be halted by wearing a mask. Edinburgh scientists thought about the quantity of drops that arrived on a surface before an individual infected and talking without and with a medical mask or an essential cotton face covering. Tests were completed on individuals and an analysis was found that the number of droplets was more than 1,000 times lower when wearing even a single layer cotton mask. However, it can be difficult to understand speech when the talker is wear- ing a mask. Most masks create a barrier during communication as they prevent visual access to the speaker’s lips. This can hinder speech perception, especially in noisy environments or when the lis- tener has a hearing impairment (Hampton et al., 2020). By studying the acoustic effects of masks on speech signals, we can learn which masks are best for speech transmission and find ways to make communication easier. Aims and Objectives The aim of the research is to examine the acoustic atten- uation caused by different face masks, including surgical masks and a N95 mask along with a face shield by carrying out a quantitative and qualitative analysis.

worm 2022

Objectives:

 To identify the sound reduction provided by each mask type by measuring the noise levels.  To identify how different age groups and genders will have a different response to the way

they perceive the speech alongside the two sound attenuation techniques used.  To interview the volunteers and attain their perception on the speech heard with the different

type of masks used.

2. LITERATURE REVIEW Face masks (alongside physical distancing) provide some protection against infection from Corona- virus disease (COVID-19). Their use in public spaces and healthcare settings is either recommended or mandatory in many jurisdictions internationally. In the United States, the Centre for Disease Con- trol (CDC, 2020) recommends mask use to minimize droplet dispersion and aerosolization of the virus. Clinical trials and healthcare settings continue to assess speech production, which generates respiratory droplets while unrestricted exposure increases the likelihood of disease contraction. Types of facemasks The most commonly type of masks used are surgical masks. These are commonly 3- ply fabric with a water-resistant outer layer, filter middle layer, and water-absorbing inner lay. Surgical masks fit loosely on the face with air able to escape from the sides. The N95 masks are similar in composition to surgical masks, however these masks have a higher filtration rate and a tight seal around the wearer’s face, preventing any air escape (O’Dowd et al., 2020). Even though it is believed that face- masks attenuate sound transmission like a low-pass filter (Coyne, et. Al), little information is availa- ble on voice characteristics whilst wearing a facemask. The infrequent literature on the topic suggests possible changes in the speech spectrum. A group of researchers compared speech spectral levels calculated as total root mean square (RMS) power from the Connected Speech Test (CST) stimuli produced by one speaker with and without wearing a surgical mask. (Mendel, et al., 2018). Recent Research Findings Given the widespread use of facemasks through the pandemic, it seems reasonable to additionally explain the qualities of voice signal in speech of vocally healthy speakers who wear a face mask. The production of clear speech is vital to accurately comprehend the speech and requires less listening effort than tainted speech. A study conducted quantified the low and high frequency energy regions which also contribute to speech recognition (Hillenbrand, J. & Houde, R. A, 1996). Low-frequency spectral bands are important in recognizing vowels and voiced fricative consonants (Hughes, G. W. & Halle, (1956). High frequency spectral energy makes a significant contribution to speech recogni- tion including the recognition of vowels, voiceless and voiced fricative consonants, spoken and sung text, and speech recognition in noise (Monson, B. B., Lotto, A. J., (2014). Additionally, high fre- quency region plays a vital role in the perception of clear speech: a shift of energy concentration toward higher frequency regions underwrites to the clear speech effect for normal-hearing listeners. Seemingly, the quality and audibility of the voice might also change whilst wearing a facemask. This

worm 2022

change may interfere with auditory-perceptual voice judgment by speech language pathologists (SLPs) and ear nose and throat specialists (ENTs) (Porcaro, C. K. et al., 2019). The pandemic has raised the stakes even higher. COVID-19 is deadlier for the elderly, who are more likely to be hard of hearing. Masks don't just muffle voices; they also filter out higher pitches. In English, these are the sounds of consonants like ss and th, which help distinguish one word from the next. Masks dampen frequencies between 2000 and 7000 Hz.

3. METHODOLOGY FRAMEWORK This project has evaluated the impact wearing a mask has on acoustic output and speech perception. By examining how different face mask types (surgical and N95 along with face shield), on a speaker affect speech recordings and perceived intelligibility. It is hypothesized that the acoustic output would be impacted based on the composition/thickness of masks while perceived intelligibility will remain largely unaffected. As masks make their way and act as a communication barrier, room acoustics play an important factor and must be considered as a critical aspect when designing spaces. Meeting rooms are spaces of communication and therefore acoustic performance is of serious im- portance. Unfortunately, when designing meeting rooms, acoustics is often not considered, and even if a designer wanted to design an amazing sounding meeting room, there exists little guidance. Due to their size and form, meeting rooms often suffer from a variety of acoustic defects including flutter echoes and room resonances. Speech and noise levels are influenced by room acoustics. Measuring sound attenuation of the face masks As a starting point for the project, a meeting room was chosen with an office located in U-Bora Towers, Dubai, UAE. Below shows the floor plan of the office fit out. Meeting Room 4 has a rever- beration time of 0.76s and is the preferred room in the office due to its location.

worm 2022

Figure 1: Office Layout.

‘(Open Plen Workspace: + Meeting Room 5 ——E Meeting Room

worm 2022

Figure 2: The meeting room with the placement setup of the microphone on the table.

Figure 3: All equipment’s used for the study.

 The audio file was then played with the 3 different setups. Each measurement was repeated

twice, and an average was taken.

Single ply Face Shield N95 face mask + shield setup RION NL-52 SLM N95 face mask

 The audio file played was a pre-recorded sentence “The hungry purple dinosaur ate the kind,

zingy fox, the jabbering crab, and the mad whale and started vending and quacking.” This sentence was chosen since it has most of the constantans and vowels. The audio clip was played via Bluetooth connection from an iPhone 11 Pro to the A10 speaker as shown in the speaker above at a volume level of 5 for all the measurements taken.  Measurements were made using a type-1 (class 0) sound level meter unit (incl ¼” calibrated

microphone) in full accordance with IEC 60651 and IEC 60804. The measuring equipment comply with Class 1, as defined in BS EN 61672:2003. The measurement system has been calibrated from an accredited laboratory (UKAS or equivalent). The next part of the study included the qualitative assessment of speech perception with the sounds played. The three audios played back were – Audio 1 – No mask Audio 2 – Surgical Mask Audio 3 – N95 Mask + Single ply Face Shield Twelve participants took part in this study – a mix of males and females. The participants were asked to hear the audio clip using their earphones at a volume level of 5 (5 scrolls up on the volume bar of the phone). The participants were English speakers, non-smokers, and only one wore hearing aids.

4. QUANTITATIVE RESULTS As per the methodology identified above, noise levels were measured with the masks placed on the speakers. For comparison’s sake, two readings were taken with one surgical mask and surgical masks on the speaker, respectively. However, there was barely any noticeable difference. In addition to masks, majority of the kiosks in the UAE use plastic protection sheets which attenuate the noise levels even further. Therefore, it should be noted that the N95 mask was clubbed along with a face shield to depict the worst-case scenario. As identified in the literature review, the sound pressure level drops at the higher frequency end for the N95 mask along with the shield. The levels for “No mask” are considerable higher towards the overall spectrum range. Table 1: Noise levels measured using the various mask types

Mask Type SPL/ dBA

No Mask 65 dBA

Surgical Mask (1 layer) 61 dBA

N95 Mask + Face shield 54 dBA

5. QUALITATIVE RESULTS

Questionnaires were given to 12 participants along with the 3 audio clips produced in mp3 format. Nearly 42% of the participants were in their 20s while just one participant was in their 50s. With

worm 2022

about 67% males and 33% females and one of the males in their 30s using hearing aids (this condition was since birth due to partial deafness), a considerable number of diverse answers was evaluated. To account for the common concerns when using face masks, questions pertaining to an individual’s experience with face masks has been highlighted in this section. 75% of the people have had to raise their voice very often while having a conversation while 50% of the people surveyed found it difficult to understand a person because they could not read their lips. Being in a diverse country like a UAE with various dialects and accents, this is a common concern amongst many. As discussed throughout this project, audio is perceived in different ways by individuals. When asked on the perception of audibility, over 83% of the participants were able to clearly hear the no mask audio clip. 41% of the participants thought it was moderately audible with the N95 and face shield masks and they had issues understanding the word “jabbering” and “venting”. The purpose of this study was to compare speech perception in different mask conditions. Overall, perception for N95 along with a face shield (Audio 3) was significantly lower compared to the no mask (Audio 1) condition. Previous studies highlighted in the literature review indicate that N95 masks attenuate sound to a greater extent than surgical or cloth masks. In the current study, percent words correctly identified was lowest for N95 masks across all mask conditions. Surgical masks (Au- dio 2) and double layer surgical masks, commonly worn during activities of daily living, did not differ from each other and showed similar effects on intelligibility. Although survey studies have reported that individuals, particularly those with hearing loss, experience difficulties with communication through masks and increased cognitive load, some studies of human speech perception to date have failed to find significant differences between masked and unmasked conditions. Taken together, these results suggest that multiple variables are relevant in determining the most appropriate mask for a given situation. There is an inverse relationship between the weave of masks (related to their breath- ability and comfort) and their ability to block respiratory droplets. These factors also interact with sound attenuation. Therefore, individuals should choose the mask types based on type of mask de- pending on a range of factors including health risks, environmental noise, communication effort, and context. 6. CONCLUSIONS

Limitations This pilot study was conducted online due to the restrictions on face-to-face data collection during the Covid-19 pandemic. Online data collection did not allow to control for differences in participants’ hardware for listening to the sentences. These results should be confirmed under more controlled laboratory conditions. Future studies will examine speech perception through masks in adults with aging-related hearing loss and other populations with communication impairments. More detailed acoustic analyses will be conducted to determine how different masks and other sources of environ- mental noise (e.g., street traffic, schools) impact production and perception of specific speech sounds. Future studies will also examine the influence of experience and attitudes toward face masks in speech perception. Conclusion

worm 2022

The two types of masks investigated in this study interfered with speech perception. The N95 mask provides the most protection from respiratory droplets but may provide the least usable information to comprehend speech in context. Critical information conveyed in healthcare settings may be sup- ported by giving patients written or visual information. Surgical masks are the most worn during activities of daily living also interfere with speech perception. Professionals in settings such as schools should consider the trade-off between clear communication and health risks when selecting which masks to use in different contexts. 7. ACKNOWLEDGEMENTS

I am firstly grateful to the Institute of Acoustics and my supervisor, as this research was a part of my IOA Diploma course. Furthermore, I’d like to acknowledge all the participants within my question- naire who supported me and my office that permitted to carry out the research in their facility. 8. REFERENCES

1. Poulsen, T. Influence of session length on judged annoyance. Journal of Sound and Vibration,

145(2) , 217–224 (1991). 2. Asadi, S., Wexler, A. S., Cappa, C. D., Barreda, S., Bouvier, N. M., and Ristenpart, W. D. (2019).

“Aerosol emission and super emission during human speech increase with voice loudness,” Sci. Rep. 9, 2348. https://doi.org/10.1038/s41598-019-38808-z 3. Atcherson, S. R. et al. The effect of conventional and transparent surgical masks on speech un-

derstanding in individuals with and without hearing loss. J. Am. Acad. Audiol. 28, 58–67 (2017). 4. CDC (2020). Use of Masks to Help Slow the Spread of COVID-19 (Centers for Disease Control

and Prevention, Atlanta, GA). 5. Corey, R. M., Jones, U., and Singer, A. C. (2020). “Acoustic effects of medical, cloth, and trans-

parent face masks on speech signals,” J. Acoust. Soc. Am. 148, 2371. https://doi.org/10.1121/10.0002279 6. Coyne, K. M. et al. Respirator performance ratings for speech intelligibility. Am. Ind. Hyg. Assoc.

J. 59, 257–260 (1998). 7. Goldin, A., Weinstein, B. & Shiman, N. How do medical masks degrade speech perception? Hear.

Rev. 27, 8–9 (2020). 8. Hampton, T., Crunkhorn, R., Lowe, N., Bhat, J., Hogg, E., Afifi, W., Krishnan, M., Street, I.,

Sujata, D., and Sharma, R. (2020). “ Speech discrimination challenges of healthcare professionals whilst wearing personal protective equipment (PPE) during the coronavirus disease 2019 (COVID-19) pandemic,” Authorea Preprints, https://doi.org/10.22541/au.159050338.83886289. 9. Hillenbrand, J. & Houde, R. A. Acoustic correlates of breathy vocal quality: Dysphonic voices

and continuous speech. J. Speech Hear. Res. 39, 311–321 (1996). 10. Mendel, . et al. Comparison of cepstral peak prominence measures using the ADSV, SpeechTool

and VoiceSauce acoustic analysis programs. Acoust. Aust. 46, 215–226 (2018). 11. Monson, B. B., Lotto, A. J. & Story, B. H. Gender and vocal production mode discrimination

using the high frequencies for speech and singing. Front. Psychol. 5, 1239 (2014). 12. O'Dowd, K., Nair, K. M., Forouzandeh, P., Mathew, S., Grant, J., Moran, R., Bartlett, J., Bird, J.,

and Pillai, S. C. (2020). “ Face masks and respirators in the fight against the COVID-19 pandemic)

worm 2022