The attenuation and speech intelligibility of the Gecko helmet used by the Royal National Lifeboat Institution Stephen Jay 1 London South Bank University 103 Borough Road London SE1 0AA ABSTRACT The Gecko helmet is used by the Royal National Lifeboat Institution (RNLI) to provide head protection for its volunteer crew members at sea, with the introduction of the more powerful Atlantic 85 lifeboat, an intercom system was introduced to aid communication between crew members on the vessel. Evidence from discussions with volunteer crew members has led to an- ecdotal evidence that the use of the helmet and intercom system reduced their ability to hear and understand unamplified voices while the helmet is fitted. The objective measurement of attenuation was undertaken within an anechoic chamber to understand how the attenuation of sound changes when the intercom was fitted. The measurement of speech intelligibility was undertaken within an anechoic chamber using STIPA signal. Subjective measurements of speech intelligibility were undertaken using 22 volunteers with a varying experience of the Gecko helmet. Subjective testing consisted of listening to 4 sets of phonetically balanced word score recordings under four conditions. Results of measurements undertaken show the helmet attenuates sound in the frequency range of 1 kHz to 16 kHz, partially effecting the key frequen- cies for speech. Objective and subjective measurements of speech intelligibility show a decrease in the intelligibility of speech when the Gecko helmet was introduced. 1. INTRODUCTION

The Royal National Lifeboat Institution (RNLI) was founded on the 4 th March 1824 with the goal to create a service dedicated to saving lives at sea (RNLI.org, 2019). From the RNLI’s beginnings to present day, the RNLI has been at the forefront of developing new and innovative equipment to help keep their volunteer crew safe at sea. With the introduction of the inshore lifeboat fleet in the 1960’s the need for head protection for their volunteer crews became apparent. Working in conjunction with Gecko helmets, the RNLI helped to create the first modern marine safety helmet that meets the safety standard PAS 028 for impact protection. The latest iteration, the Mark 11 (shown in Figure 1) incorporates an inflatable head cushioning to allow for the correct fitting of the helmet to most head sizes and removable intercom and microphone. With the incorporation of helmets to provide head protection, the helmet design now covers the Pinna and lower skull with circular cut outs at the location to allow a direct transmission path to the ear. With the introduction of the intercom system and the use of a full-face visor sound transmission paths

1 sjaygbr@googlemail.com

to the ear are reduced. With the inclusion of masking noise from engine noise and to a lesser affect wind and wave action, this has led to a loss of external noise perception leading to issues with com- munication.

Figure 1. The RNLI Mk 11 Gecko Helmet 2. METHODOLOGY 2.1 Measurement of attenuation on the Gecko Helmet Mlyinski et al (2009) and Cespón et al (2013) have previously undertaken similar experimen- tsinvestigating the effect of noise attenuation from motorcycle helmets, where the noise source is set at a known distance and angle of incidence to simulate noise sources from a motorcycle. By broadly following a similar test method, the effect of noise on the Gecko helmet can be replicated within an Anechoic Chamber. The main noise source RNLI volunteers are exposed to on the Atlantic 85 comes from the two 115 horse power outboard engines located at the stern of the Atlantic 85 lifeboat. With the location of the engines to the stern, the volunteer crew member sat on the fourth man seat on the Atlantic 85 would be considered as the worst-case position for noise exposure. Mlyinski et al and Cespón et al had differing opinions on the attenuation of motorcycle helmets when worn. Cespón et al concluded only a 12 dB of attenuation was seen above 800 Hz, while Mlyinski et al found a mo- torcycle helmet attenuated sound between 30-35 dB between 500 Hz and 8000 Hz. Although the methodology can be compared to that proposed, the type of helmet differs considerably between a motorcycle helmet and the RNLI Gecko helmet. The main difference being the Gecko helmet con- tains no polystyrene impact protection and is open faced, so reducing the ability to attenuate sound (Gecko Head Gear: 2019). Based on the worst-case exposure location on the Atlantic 85 lifeboats, a speaker was positioned with the same angle of incidence and distance to replicate the sound level from the 2 outboard engines.

a »

Following experimental methodology set out by Cespón et al (2013) a head and torso simulator (HATS) was used to imitate the crew position and allow for ease of replication of the conditions of the experiment. Measurements were undertaken within an anechoic chamber to replicate the free field environment as would be the normal condition on the Atlantic 85 lifeboats. 2.1 Speech Intelligibility The test methodology for speech intelligibility has been split into two separate methods: subjec- tive and objective experiments to fully understand the effect of the helmet. Objective measurement of speech intelligibility was undertaken using a speech transmission index for public address systems (STIPA) signal. With the use of a STIPA signal under free field condi- tions, an understanding of the effect of the angle of incidence can be undertaken without overbur- dening test volunteers. Measurements were undertaken using a Brüel and Kjær HATS system (Model 4100), with the Gecko helmet fitted in two states, helmet on with intercom not fitted and helmet fitted with intercom installed. A third test was undertaken with the helmet removed to pro- vide baseline measurements to allow a comparison to be undertaken. Measurements were under- taken with no masking noise, engine idling noise and pink noise. The STIPA signal was reduced in increments of 5 dB from a maximum level of L Aeq 75 dB to L Aeq 55 dB, this reduction in STIPA signal in relation to the level of the masking noise will allow an understanding of how the signal to noise ratio effects speech intelligibility when the Gecko helmet is fitted. A STIPA signal was gener- ated at a distance of 2 metres from the HATS system with an angle of incidence of 0º and 90º to the HATS system with 0º denoting that the speaker is facing the front of the HATS system. Subjective measurement of speech intelligibility was undertaken using 22 volunteers both experi- enced and inexperienced in the wearing of the Gecko helmet. Testing was undertaken using four sets of phonetically balanced word lists containing 50 words played through a Genelec 8020 bi-am- plified monitor system using a Focusrite Scarlett 6i6 amplifier using Audacity open source audio software to control level and gain during the experiment. The level of amplified speech was set at approximately L Aeq 60 dB at 1.5 metres, with the actual level dependent on the words spoken within the phonetically balanced word list. A second speaker was set up behind the test volunteer at a dis- tance of 1.4 metres to imitate the location of the dominant noise source on the Atlantic 85. The sound level of the outboard engines idling had previously been measured at the approximate head position of the fourth man seat on the Atlantic 85 and was recorded at L Aeq 64 dB, this was repli- cated using a looped recording of the Atlantic 85’s engines idling. Volunteers were placed in a sit- ting position to replicate the fourth man position on the Atlantic 85 with volunteers asked to listen to the four sets of phonetically balanced word sets in four states, no helmet and no masking noise, no helmet and masking noise, helmet with no masking noise and helmet with masking noise. Tests were undertaken with intercom fitted only, this was due to time constraints to undertake subjective measurements. All volunteers were asked to complete a short survey before undertaking the experi- ment providing data on age, previous experience of wearing the Gecko helmet, and known hearing impairments with all tests anonymized and linked via test number for comparison. BS EN 60268-16:2020 shows the relationship between phonetically balanced word scores and speech transmission index, this has been replicated in Figure 3.

Figure 2. Phonetically balanced word score relationship to speech transmission index (BS EN 60268-16:2011) Figure 2 shows that the relationship between phonetically balanced word scores and STI is not a linear relationship, as such a STI result above 0.75 can be compared to a Phonetically based word score test result of over 95% intelligibility. 3.0 RESULTS 3.1 Attenuation of the Gecko Helmet Table 1 sets out an overview of measurements undertaken during objective sound attenuation measurements undertaken at the left and right ear positions, the results reported are the average of measurements undertaken at all angles of incidence measured. Overall results show that the Gecko helmet provides limited attenuation when fitted to the HATS system. Figure 4 shows the attenuation of the Gecko helmet between 6.3 Hz and 20 kHz at the left and right ear positions at 0º angle of incidence.

Noise Source Helmet State Left Ear Right Ear Log average L Aeq dB

Highest L Max dB

Log average L Aeq dB

Highest L Max dB Background No Helmet 16.2 22.7 17.4 30.8 Helmet, no intercom 16.7 30.1 19.5 43.1 Helmet and Intercom 16.2 22.1 18.1 40.3 Engine Idling No Helmet 62.6 67.4 68.4 73.2 Helmet, no intercom 66.8 71.7 69.5 74.6 Helmet and Intercom 66.2 71.8 66.4 72.1 Boat Speed 21

Knots No Helmet 85.6 91.1 88.4 91.6 Helmet, no intercom 85.7 89.6 88.9 91.8 Helmet and Intercom 85.4 88.3 85.5 88.7

No Helmet 99.6 104.1 102.9 105.6 Helmet, no intercom 100 103.2 103.2 106.7 Helmet and Intercom 103 102.9 101 104.5

Boat Speed 35 Knots

Pink Noise No Helmet 109 113.7 112.5 115.8 Helmet, no intercom 110 112.9 113.1 166.6 Helmet and Intercom 111 113.7 113.1 117 Table 1. Overview of attenuation measurements Overall results show that the Gecko helmet provides limited attenuation when fitted to the HATS system. Figure 4 shows the attenuation of the Gecko helmet between 6.3 Hz and 20 kHz at the left and right ear positions at 0º angle of incidence. The Gecko helmet attenuates sound at levels above 1 kHz to at both the left and right ear positions with the helmet fitted with the intercom. Measure- ments are the average result of two helmets tested with a minimum of five measurements per hel- met. Results differ slightly between left and right ears. When the helmet was fitted to the HATS system without an intercom system, attenuation began at approximately 2 kHz. The difference measured in the attenuation from the baseline measurements show a similar level of attenuation between the helmet fitted, with and without the intercom system. Five measurements per helmet state were undertaken with the helmet not being removed between measurements.

0 20 40 60 80 100 120

Sound pressure level (dB)

6.3 Hz 8.0 Hz 10 Hz 12.5 Hz 16 Hz 20 Hz 25 Hz 31.5 Hz 40 Hz 50 Hz 63 Hz 80 Hz 100 Hz 125 Hz 160 Hz 200 Hz 250 Hz 315 Hz 400 Hz 500 Hz 630 Hz 800 Hz 1.0 kHz 1.25 kHz 1.6 kHz 2.0 kHz 2.5 kHz 3.15 kHz 4.0 kHz 5.0 kHz 6.3 kHz 8.0 kHz 10.0 kHz 12.5 kHz 16.0 kHz 20.0 kHz

1/3 Octave frequency band

No Helmet, left ear Helmet, no intercom left ear Helmet, intercom fitted left ear No Helmet, right ear Helmet, no intercom right ear Helmet, intercom fitted right ear

Figure 3. Pink noise attenuation of the Gecko helmet 3.2 Speech Intelligibility The average result of objective measurements of the speech intelligibility of the Gecko helmets tested is shown in Table 2.

0°Angle of incidence No masking Noise Engine Idling (65 dB) Pink Noise 65 dB STIPA Signal level L R L R L R

75 dB 0.89 0.85 0.83 0.77 0.82 0.74 70 dB 0.90 0.86 0.84 0.78 0.72 0.64 65 dB 0.90 0.88 0.67 0.62 0.60 0.52 60 dB 0.91 0.87 0.63 0.50 0.45 0.38 55 dB 0.88 0.87 0.44 0.41 0.30 0.25

Intercom fit- ted

75 dB 0.86 0.85 0.78 0.76 0.76 0.76 70 dB 0.86 0.86 0.71 0.70 0.66 0.65 65 dB 0.89 0.88 0.58 0.58 0.52 0.51 60 dB 0.89 0.87 0.44 0.45 0.35 0.34 55 dB 0.90 0.88 0.37 0.23 0.26 0.26 Table 2. Averaged speech transmission index results of all helmets tested Results show that when the helmet was fitted with and without an intercom system a decrease in the STI result was seen. What had not been considered, but is shown in the measurements is that with no intercom fitted (Pinna exposed) the STI score was lower than the with the intercom fitted (Pinna

No intercom

covered). With the inclusion of masking noise, the STI score reduced when the difference in masking to STIPA signal ratio was by 10 dB. Although reduced, the STI score can be considered to be rated A+ based on BS EN 60268-16:2011 nominal classification. The trend of a lower STI score with no intercom fitted continued with the reduction in level of the STIPA signal in relation to the masking noise level. Figure 4 shows the results of subjective testing of speech intelligibility of the Gecko helmet based on age groups. Results show that the introduction of the helmet leads to a decrease in performance in scores in relation to scores when the helmet was not worn.

Figure 4. Phonetically based word score by age group A comparison of objective and subjective test results can be made based on PB word score per- centage intelligibility and quality according to BS EN 60268-16:2011. Using the comparison table within BS EN 60268 the average results of both objective and subjective tests were compared and is shown in table 3. For the purpose of comparison, a signal level of 60 dB has been selected as this is considered to mimic the level of normal speech sound pressure levels. Table 3 shows that results from objective measurements using a STIPA signal achieved a higher result on the common intelli- gibility scale with both the helmets fitted. Objective test methods achieved a score of excellent in comparison to a score of good by subjective test methods. With the introduction of masking noise using replicated engine idling noise, results still vary between objective and subjective measure- ments. Measurements using a STIPA signal achieved a fair rating, while subjective testing achieved a rating of bad, the poorest result achievable.

peare

Engine idling

No masking noise

masking noise

Objective measurement No helmet Excellent Poor

Helmet and intercom Excellent Fair

No helmet Good Poor

Subjective measurement

Helmet and intercom Fair Bad

Table 3. Subjective and objective common intelligibility scale results 4. CONCLUSIONS The RNLI Gecko helmet is worn by all inshore lifeboat crew to protect against head injuries, with the introduction of the Atlantic 85 an intercom system has been fitted to the helmet, blocking the two openings that allowed direct passage of sound to the ear. Since its introduction it has been noted that the ability to understand conversations at normal speech levels has decreased. Objective measurements of attenuation and speech intelligibility was undertaken using the Gecko helmet within an anechoic chamber, results showed that the Gecko helmet only provided limited attenuation of sound at frequencies above 1 kHz. Speech intelligibility measurements were undertaken using both objectively and subjectively methods. A comparison of results undertaken. STI measurements of the Gecko helmet showed that with the helmet fitted and the inclusion of masking noise, the STI score reduced more then based on the helmet alone. Subjective measurements showed the same correlation as the objective measure- ments, where with the introduction of the Gecko helmet with the intercom system fitted, speech in- telligibility was reduced. 5. REFERENCES 1. BS EN 60268-16:2011: Sound system equipment. Objective rating of speech intelligibility by speech transmission index, (2011) Sep 30. Available from: https://bsol.bsigroup.com/en/Bsol- item-Detail-Page/?pid=0000000 000 30249993 . 2. Cespón, J. L. G., Vilán, J. A. V., Seoane, M. A. S. and Guijarro, S. T. (2013) Characterization of the attenuation properties in motorcycle helmets, Acta Acustica United with Acustica , 99, pp. 1008-1013. DOI: 10.3813/AAA.918679. 3. Gecko Head Gear. Manufacturing Helmets For Lifeboat Crews Nationwide, Gecko Head Gear ,. Available from: https://www.geckoheadgear.com/sectors/lifeboat-crews-2/ [Accessed Sep 9, 2019]. 4. Mlynski, R., Kozlowski, E. and Zera, J. (2009) Attenuation of noise by motorcycle safety hel- mets, International journal of Occupational Safety and Ergonomics , 15, pp. 287-293. Available from:https://search.ebscohost.com/login.aspx?di- rect=true&db=edswss&AN=000269853300005&site=eds-live. 5. Timeline – Our History. Available from: https://rnli.org/about-us/our-history/timeline [Ac- cessed Sep 9, 2019].