Influence of transient pressure changes on speech intelligibility: Implications for next-generation train travel.

High-speed trains are operated in increasingly complex railway networks and continual improvement of driver assistance systems is necessary to maintain safety. Speech offers the opportunity to provide information to the driver without disrupting visual attention. However, it is not known whether the transient pressure changes inside trains passing through tunnels interfere with speech intelligibility. Our primary goal was to test whether the most severe pressure variations occurring in high-speed trains (25 hPa in 2 s) affect speech intelligibility in individuals with normal hearing ability and secondly whether a potential effect would depend on the direction of the pressure change. A cross-over design was used to compare speech intelligibility, measured with the monosyllable word test by Wallenberg and Kollmeier, in steady ambient pressure versus subsequent to pressure events, both realised in a pressure chamber. Since data for a power calculation did not exist, we conducted a pilot study with 20 participants to estimate variance of intra-individual differences. The upper 80% confidence limit guided sample size of the main campaign, which was performed with 72 participants to identify a 10% difference while limiting alpha (5%) and beta error (10%). On average, a participant understood 0.7 fewer words following a pressure change event compared to listening in steady ambient pressure. However, this intra-individual differences varied strongly between participants, standard deviation (SD) ± 4.5 words, resulting in a negligible effect size of 0.1 and the Wilcoxon signed rank test (Z = -1.26; p = 0.21) did not distinguish it from chance. When comparing decreasing and increasing pressure events an average of 0.2 fewer words were understood (± 3.9 SD). The most severe pressure changes expected to occur in high-speed trains passing through tunnels do not interfere with speech intelligibility and are in itself not a risk factor for loss of verbal information transmission.

1 Introduction

Modern societies are characterised by a growing demand for mobility of individuals and high-speed trains are increasingly becoming a backbone of the transportation infrastructure in many countries [1]. To facilitate safe operation of these fast trains in increasingly complex railway networks a multitude of driver assistance systems are currently under development [2]. In human-machine interfaces the auditory channel can be used to provide information to the user, without disrupting visual attention [3]. This is particularly advantageous when operating heavy mobile machinery, such as trains. Speech has hereby an advantage over other acoustic stimuli; it does not only alert the operator but it carries the relevant information itself. However, if speech is considered for transmission of essential information in an operational environment it must be ensured that the acoustical transmission path is interruption-free. Therefore it is necessary to assess the train environment for interference with speech intelligibility.

Clearly, speech can only be understood, if it is heard. A basic metric used to describe audibility is therefore the sound pressure level (SPL) of the speech signal relative to the background noise, i.e. signal-to-noise ratio (SNR). A rule of thumb suggests an SNR of 6 dB(A) is required for minimal and 10 dB(A) for good speech intelligibility [4]. Beyond SPL, spectral composition of speech and noise also play an important role [5], e.g. the more noise intersperses into the speech frequency band the stronger its disruptive impact. More advanced metrics, such as the Speech Transmission Index [6], the Articulation Index [7] and the Speech Intelligibility Index [8] incorporate additional acoustic properties to deduce transmission quality and the amount of speech reaching the listener. But while the influence of acoustic parameters is well understood, little attention has been paid to other environmental factors, such as ambient pressure, humidity or temperature [9].

The fast transient pressure changes experienced in high-speed trains when passing through tunnels [10] are associated with aural discomfort [11], can attenuate sound transmission through the outer and middle ear [12] and have thus the potential to interfere with speech intelligibility. Repeated tunnel passing has even been linked to persistent hearing loss in train drivers [13], though the primary concern here was tunnel noise. In general, negative middle ear pressure is a common pathological dysfunction and has been intensely studied [14]. It is known that static pressure gradients across the tympanic membrane impair hearing sensitivity [15] and speech intelligibility has been looked at as a function of altitude in aircraft [16]. However, the long and slow pressure gradients in airplanes provide opportunity to equilibrate middle ear pressure and induce hypobaric hypoxia with increasing altitude. To our knowledge, no data is available to discern whether the small and rapid transient pressure variations occurring inside high-speed trains passing through tunnels can affect intelligibility of speech. Consequently, it is not known whether such variations may be a source of interference when using speech to indicate potentially critical information to train drivers.

Our primary objective was to test whether transient pressure changes of 25 hPa in 2 s, the most severe pressure events expected to occur in high-speed trains [10], affect speech intelligibility in individuals with normal hearing ability. Our secondary question was to assess whether the direction of the pressure change, i.e. increasing or decreasing ambient pressure, affects understanding of speech differently. We used the monosyllabic rhyme test devised by Wallenberg and Kollmeier (WAKO) to measure speech intelligibility [17], which is based on understanding of individual words.

2 Material and methods

This prospective study was approved by the ethical committee of the medical association North Rhine-Westphalia (approval date June 6, 2017; number 2017121) and performed at the German Aerospace Centre (DLR) in Cologne (Germany). The pilot study for sample size estimation was performed in August 2017 and the main data collection took place between September and October 2017.

2.1 Sample size estimation

It was our aim to conclusively answer the question as to whether or not speech intelligibility is affected by transient pressure changes. Since statistical tests only control for false positive results it was crucial to study a sufficiently sized sample in order to restrict false negative findings as well. We could not identify any published data enabling a reliable power calculation; therefore we conducted a pilot study with 20 participants (10 female, mean age 27 years ± 6 SD). Variability estimates from pilot data tend to underestimate the population parameter [18]. To account for this we calculated the upper 80% confidence limit for the standard deviation (SD) of intra-individual differences (7.45 words) and used it to guide sample size of the main campaign [19]. This resulted in recruitment of 72 participants, enabling us to identify a 10% difference in speech intelligibility within an individual, while limiting the chance of alpha and beta error to 5% and 10% respectively. Rounding the sample size up to the next multiple of 8 was required to enable permutation of experimental conditions.

2.2 Participants

Both female and male individuals with unimpaired hearing ability were eligible for the experiment. Out of the 72 volunteers, aged 19 to 39 years (mean 25 ± 4 SD), participating in the main campaign 35 were female. Pilot study participants were not included in the main trial. Participants were recruited from a database of former study attenders at the DLR Institute of Aerospace Medicine and spoke German as their native language. Normal hearing ability, defined as no more than 20 dB reduction in any frequency band of 500 Hz and 1000 to 4000 Hz (in 1000 Hz steps), middle ear pressure and mobility of the eardrum were ensured by audiometry (Device AD226, Interacoustics, Audiometer Alle, 5500 Middelfart, Denmark) and tympanometry (Device TITAN, also Interacoustics), both measured before and after the experimental procedure in the pressure chamber. All volunteers gave written informed consent before starting the test protocol and were in good health as assessed by general questionnaires. Participation was reimbursed with 90 Euro.

2.3 Word test

Speech intelligibility was measured using the monosyllable word test by Wallenberg and Kollmeier (WAKO) [17], commercially distributed by HörTech gGmbH, Oldenburg, Germany. The test uses a closed-set response method; subsequent to each audibly presented test word the participant is asked to identify the word from five written alternatives. Each alternative differs in one of three phonemes. The test can be used in quiet and in noise and since it is based on the understanding of individual words it qualifies to discern the effect of short recurrent events, like changes in ambient pressure.

2.4 Experimental design

The pressure chamber used for the experiment is a former deep diving facility, essentially a horizontally oriented metal cylinder, capable of producing rapid and highly accurate pressure variations [11]. The barrel-shape of the facility is well suited to create homogeneous noise environments and has been used for this purpose in previous studies simulating train environments [20]. We used four loudspeakers for playback of the background noise, two at both ends of the cylinder, each facing the round chamber wall. In addition, two separate loudspeakers, installed at the round ceiling, were used for announcement of the test words. The sound at each seat was calibrated using a microphone at ear height. A schematic of the test set-up is shown in Fig 1.

Fig 1

Experimental assembly in the pressure chamber.

A real time controller synchronised chamber pressure, announcement of WAKO test words and response collection via tablet computers. A homogeneous acoustic environment was created using independent signal processors for each of the four loudspeakers (LS) for noise and the two LS for test word playback.

We used a cross-over design in which each participant was studied in a single 1 hour session. Participants were examined in groups of four and each session started at 1 o’clock in the afternoon with a familiarisation to the speech intelligibility task using the same 10 test words for all participants. Afterwards, participants were presented with two test blocks of 50 words each. In one block, each test word was played without any delay after a linear change in ambient pressure of 25 hPa within 2 s, while the other block was performed at steady ambient pressure of 950 hPa. To account for potential order effects (e.g. due to time on task), we permuted the experimental conditions, i.e. testing with and without pressure changes, while maintaining the order of words throughout each test block. To call attention to the upcoming test word each presentation was preceded by an audible announcement (“please mark the word”). Each word was followed by a 20 second break during which participants selected the answer on a tablet computer and had the opportunity to equilibrate middle ear pressure. Instead of repetitively presenting pairs of increasing and decreasing pressure we constructed an irregular sequence of positive and negative pressure changes. This prevented participants from anticipating the direction of the upcoming pressure change. The sequence was balanced in regard to increasing and decreasing events and the number of successions in each direction. The minimum chamber pressure was restricted to 900 hPa in order to avoid onset of hypoxia. A speech simulating background noise (recommended spectral composition for WAKO) of 67 dB(A) was constantly played throughout the experiment, simulating the acoustic surrounding of a train and masking the operating sounds of the chamber. Prior to our experiment the direction of the effect was unknown, i.e. it was not clear whether pressure changes would increase or decrease speech intelligibility and it was possible that pressure changes in different directions could have opposite effects. Thus, the experiment needed to be equally sensitive in both directions. This was achieved by calibrating SPL of the presented test words to yield 50% speech intelligibility in steady ambient pressure, resulting in an SNR of approximately 0 dB, i.e. speech and noise signal had nearly the same SPL.

2.5 Data analysis

R version 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria) was used to perform all calculations. We planned to test both our hypotheses using a two-sided Wilcoxon signed-rank test with Pratts extension to handle ties [21], since we expected the data to be symmetrically but not normally distributed. To maintain an overall family-wise error rate of 5% we devised hierarchical testing using complete alpha spending from the primary to the secondary hypothesis, i.e. only testing for an effect of direction if we observed an overall effect of pressure changes on the number of correctly identified words within participants. After we had defined our a priori analysis protocol a simulation study of paired count data was published [22], attesting that we had chosen an appropriate and unbiased strategy, but that a paired t-test could, in comparison, yield slightly improved power, despite the data not being compliant with its test assumptions. This encouraged us to present standard deviations and t-distribution based confidence intervals of our results.

Since we expected missing data to occur unsystematically and only occasionally as a result of attention lapses, we did not plan an imputation strategy. Participants would immediately be notified in case they missed a response, bringing their attention back to the task. We recorded 7194 of the nominal 7200 responses, the six missing replies originated from five different participants and three occurred in each experimental condition. The assumption of these being missing completely at random appears justified.

3 Results

In the condition without pressure variations the average number of correctly identified words per participant was 29.9 (± 3.4 SD) out of 50, i.e. 49.7% (± 8.5 SD) of words were truly understood when considering each trial had a 20% chance for correct guessing. This shows that the targeted calibration of a 50% intelligibility rate at steady ambient pressure was achieved within reason.

When comparing the number of correctly understood test words announced immediately following a pressure change event to the number correctly understood in steady ambient pressure within each participant, the average understanding was 0.7 fewer words in the former condition. Due to the high variability of this intra-individual differences between the study participants, it has a standard deviation (SD) of ± 4.5 words, this equates only to a negligible effect size of 0.1 and the Wilcoxon signed rank test (Z = -1.26; p = 0.21) does not detect it to be distinguishable from chance level. When comparing decreasing and increasing pressure events within each participant the average understanding differed by 0.2 words (± 3.9 SD), but this was not formally tested due to our hierarchically ordered hypothesis and lack of statistical significance of the primary hypothesis. The corresponding confidence intervals are displayed in Fig 2. Visibly, no difference, i.e. zero, is within the interval borders.

Fig 2

Estimators and 95%-confidence intervals.

Intra-individual differences in number of words understood, CI based on t-distribution (71 degrees of freedom).

4 Discussion

Disequilibrium of tympanic air pressure has been reported to promote conductive hearing loss by changing the mechanical properties of the tympanum [23]. Thus, it has not been farfetched to consider pressure variations in trains passing through tunnels a risk factor for verbal information loss. To close this knowledge gap we studied the phenomenon using a highly controlled experimental setup and examined both positive and negative variations in ambient pressure. The results of this study indicate that speech intelligibility is not impaired by transient pressure changes in the order of magnitude expected during train travel.

We studied four participants during each test session and used an established calibration procedure to assure the quality of the acoustic environment [20]. But since the calibration was performed without having participants, i.e. four human bodies, inside the chamber, it is still possible that the acoustic properties at the different seats may have varied slightly during the actual experiments. However, the within-subject design, in combination with the fact that participants did not change seats during the procedure, ensures that the intra-individual differences in words understood are an unbiased representation of the effect of pressure changes on speech intelligibility.

We chose amplitude (25 hPa) and duration (2 s) of our intervention to slightly exceed the severity of events that occupants experience aboard current high-speed trains [10]. The positive dose-response relationship between aural impairment and severity of pressure events [15] enables to generalise our findings to scenarios of lesser severity, essentially covering the complete range of expectable pressure events in trains. Due to the level of discomfort associated with pressure variations it is very unlikely that current limits for such events in trains will change in the future [11]. Since we considered it equally important to quantify our confidence in positive as well as negative findings, we paid close attention to selecting an appropriately sized sample [18]. This secures the validity of our negative result, i.e. no difference between the two experimental conditions, with 90% certainty.

Prior studies have shown changes in aural perception by negative middle ear pressure, particularly in the speech band [15]. This is considered to have a degrading effect on the understanding of speech and is known to impair language acquisition in children with this condition [24]. It has been hypothesised that positive pressure in the ear canal would cause analogous effects [14]. At first glance, this appears contradictory to our observation of unimpaired speech intelligibility subsequent to bidirectional pressure variations, but an important difference must be taken into account: while previous physiological studies were carried out in quiet, our study measured understanding of speech in the presence of background noise, similar to what is experienced during actual train travel [20]. From a mechanistic point of view, pressure gradients across the tympanic membrane will lead to changes in the transfer function of the hearing organ [25], i.e. frequencies across the hearing spectrum are differently attenuated. This may reduce the power of relevant components in the speech signal, which in turn reduces its intelligibility [5]. However, when listening to speech at sufficient SPL in the presence of background noise the main problem for understanding becomes the overlay of the polluting signal [4]. In the noisy train environment any attenuation due to un-equilibrated ear pressure will, assuming a similar frequency composition, affect the speech and the noise signal equally and consequently not change the overall SNR. The test words in the experiment were announced with approximately the same SPL as the background noise playback, i.e. 67 dB(A). It appears reasonable to assume that, despite frequency attenuation due to ear pressure, all characteristics relevant for speech intelligibility will have remained well above the hearing threshold. In essence, we speculate that we did not observe a difference in speech intelligibility between our experimental conditions since the pressure events did not significantly change SNR or audibility of the speech signal, giving participants in both conditions approximately 50% chance to understand the test words.

While our results imply that pressure variations in trains caused by tunnels are unproblematic for using speech as a communication modality in driver assistance systems, the tunnel noise may still pose a considerable obstacle for this application. For good speech intelligibility a suitable SNR, approximately 10 dB(A) [4], must be ensured under any circumstances but at the same time peak volume of announcements must be limited to prevent noise stress and its detrimental short and long term consequences [26]. Thus, only if the operational environment is sufficiently quieted, e.g. by proper insulation/sound-attenuation of the vehicle body, speech based information transmission will be a safe tool.

5 Conclusion

High-speed train drivers are not only operating their vehicles in ever more complex railway infrastructures, but they are also subject to a multitude of factors, such as monotonous operating cycles and shift work, known to increase the likelihood of fatigue and human error [27], consequences of which may be catastrophic. Advanced assistance systems are a necessity to support train drivers and ensure safe rail operation in this ever more demanding environment. The present study found that fast transient pressure changes in itself did not interfere with speech intelligibility and therefore do not appear to be a risk factor for disruption of verbal communication. This suggests that driver assistance systems can safely make use of speech to communicate relevant information. However, speech intelligibility requires sufficient SNR and a limitation of the overall sound pressure level. The train driver must therefore operate in a sufficiently quiet environment. While this may be achieved by acoustic-insulation of the train body, it is highly desirable to aim for an overall reduction of noise emission by the train, since railway noise is not only a relevant issue inside the cabin, but also an overarching problem for residents in proximity of railway lines [28], causing annoyance and sleep disturbance [29].

(PDF)

Click here for additional data file.

13 Feb 2020

PONE-D-19-26567

Influence of transient pressure changes on speech intelligibility: implications for next-generation train travel

PLOS ONE

Dear Mr. Rooney,

Thank you for submitting your manuscript to PLOS ONE.

The manuscript has been assessed by two reviewers; their comments are available below.

While one of the reviewers has provided a positive assessment, the other has noted major concerns about aspects of the study design and notes that further experimental data are required to support the conclusions drawn.

We ask that you revise the manuscript to carefully address the points raised during the review process.

Please note that the revised manuscript will need to undergo further review, we thus cannot at this point anticipate the outcome of the evaluation process.

We would appreciate receiving your revised manuscript by Mar 29 2020 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Iratxe Puebla

Deputy Editor-in-Chief, PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: No

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: No

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This study examined the effects of transient changes in atmospheric pressure on speech intelligibility in travel trains. There has been little research on the effects of environmental factors such as ambient pressure on speech intelligibility. Therefore, the originality of this study is very high and I think it is an interesting research topic.

However, the scope of the experiment designed in this study is very limited, and complementary is essential for the experimental methodology. A study by Bostron et al. (2011) found that atmospheric effects in airplanes are small at low or high SNR but significant at moderate SNR. However, this study only evaluates speech intelligibility for the same sound pressure level (SNR = 0 dB) as background noise, and therefore has a clear limitation in the experimental design. In addition, although the direction of the speaker that reproduces noise in the experimental environment is toward the subject, it is considered that a sufficient diffused sound field will be realized only when the direction of the speaker faces the wall. If not, the speaker setup used by the author needs to be compensated for because of the directional influence of the sound source.

Therefore, this paper is not suitable for publication on PLOS ONE.

[9] Bostron JH, Brungart TA, Barnard AR, McDevitt TE. Atmospheric effects on voice command intelligibility from acoustic hail and warning devices. The Journal of the Acoustical Society of America. 2011;129(4):2237-44. Epub 2011/04/12. doi: 10.1121/1.3559710. PubMed PMID: 21476678.

Reviewer #2: The paper report a study about intelligibility in trains due to changes in pressure, performed by means of measurements and questionnaires. The study is correctly performed, but no significant results are obtained. This will not be a paper that can leave the mark, but is still publishable.

Minor correction are reported to the authors.

• I suggest to avoid the abstract style used, which is too analytic. It is important to report what are the background, objective, methods, and so on, but it is not important to read them explicitly written.

• A result of 0.7 plus or less 4.5 is a very bad number in math. It is simply very bad to read and should be reported in some different ways.

• Conclusions are a bit short and should summarize better the paper. Furthermore, I suggest, and kindly ask, to mention at the end a sentence giving attention to the very important topic of railway noise impact over citizens, which is still very annoying and deserving attention and prevention. “However, railway noise still remains an issue for people living in the nearby of railway axis, deserving community attention and careful planning in order to avoid citizens’ complaints and annoyance [Licitra, G., Fredianelli, L., Petri, D., & Vigotti, M. A. (2016). Annoyance evaluation due to overall railway noise and vibration in Pisa urban areas. Science of the total environment, 568, 1315-1325; Bunn, F., & Zannin, P. H. T. (2016). Assessment of railway noise in an urban setting. Applied Acoustics, 104, 16-23.].

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

29 Mar 2020

Answers to comments from Reviewer 1

This study examined the effects of transient changes in atmospheric pressure on speech intelligibility in travel trains. There has been little research on the effects of environmental factors such as ambient pressure on speech intelligibility. Therefore, the originality of this study is very high and I think it is an interesting research topic.

Question 1: However, the scope of the experiment designed in this study is very limited, and complementary is essential for the experimental methodology. A study by Bostron et al. (2011) found that atmospheric effects in airplanes are small at low or high SNR but significant at moderate SNR. However, this study only evaluates speech intelligibility for the same sound pressure level (SNR = 0 dB) as background noise, and therefore has a clear limitation in the experimental design.

The study of Bostron et al. examines propagation of sound in different atmospheric conditions, i.e. combinations of pressure, humidity and temperature. We cited the afore-mentioned study in our initial manuscript and have now stated their findings more explicit in the methods section, since these are crucial for the experimental setup we chose. Bostron et al. studied intelligibility of voice commands from acoustic warning devices, travelling over different distances on an airfield (receivers were positioned in 7 locations between 200m and 1500m away from the sender) at different days (i.e. different sets of atmospheric parameters, determined by the meteorological conditions of that day). The signal to noise ratio (SNR) at the different receiver positions was one of the explanatory variables they measured. They found that sound pressure levels (SPL) did not only decrease with distance, but were also affected by the meteorological conditions. They studied different SNR conditions by varying SPL of the voice commands and found that atmospheric conditions did only affect intelligibility of the voice commands at moderate SNR. At both high and low SNR, intelligibility of the voice commands was nearly independent of the atmospheric conditions, i.e. in the former voice commands were always understood and in the latter they were never understood.

Since we intended to study the isolated effect of pressure changes on speech intelligibility (not the effect of different absolute pressure levels as Bostron et al. did) this observation is pivotal for our study design and it is the particular reason why we exclusively measured at SNR = 0 dB. Prior to our experiment the direction of the effect of transient pressure changes was unknown, i.e. it was not clear whether pressure changes would increase or decrease speech intelligibility and it was also possible that pressure changes in different directions (i.e. increasing or decreasing pressure) could have opposite effects. Thus, our experiment needed to be equally sensitive in both directions. When the monosyllable word test by Wallenberg and Kollmeier is done at 0 dB SNR it provides a 50% chance for correct understanding of the test words, i.e. participants should correctly identify half of the test words (after removing the 1 in 5 chance for correct guessing). We checked the realisation of this assumption post-hoc by looking at the results of the steady ambient pressure condition only and found 49.7% of test words were identified correctly. By having the test in this “middle position”, our experiment was equally sensitive for increases as well as decreases in speech intelligibility and this was what we aimed for. Studying other SNR conditions would have provided information on the effect of SNR on speech intelligibility, which was not the goal of our experiment. Adding additional SNR scenarios would have increased the length of the experimental procedure beyond the attention span known to be acceptable for participants in this type of experiment and it would have increased the likelihood of floor and/or ceiling effects. To avoid these two problems we studied exclusively at an SNR of 0 dB, since this was optimal in the scope of our experiment.

To illustrate this situation we added the following sentence to the methods section:

Prior to our experiment the direction of the effect was unknown, i.e. it was not clear whether pressure changes would increase or decrease speech intelligibility and it was possible that pressure changes in different directions could have opposite effects. Thus, the experiment needed to be equally sensitive in both directions. This was achieved by calibrating SPL of the presented test words to yield 50% speech intelligibility in steady ambient pressure, resulting in an SNR of approximately 0 dB, i.e. speech and noise signal had nearly the same SPL.

[9] Bostron JH, Brungart TA, Barnard AR, McDevitt TE. Atmospheric effects on voice command intelligibility from acoustic hail and warning devices. The Journal of the Acoustical Society of America. 2011;129(4):2237-44. Epub 2011/04/12. doi: 10.1121/1.3559710. PubMed PMID: 21476678.

The indicated scenario of speech intelligibility as a function of altitude, i.e. pressure, in airplanes was studied by Wagstaff et al. (1999). However, the slow and gradual pressure gradients in airplanes do not compare to the rapid pressure variations experienced in the train scenario studied here. The former provides opportunity to equilibrate in ear pressure and induces hypobaric hypoxia with increasing altitude, while the latter scenario prevents pressure equilibration due to the brevity of the events and does not induce hypoxia. Wagstaff et al. looked at the effects of different constants pressure levels on hearing, while the novelty of our study is that it investigates the immediate effect of small but rapid pressure events (without inducing hypoxia) on speech intelligibility.

We added sentences to this effect to the introduction:

It is known that static pressure gradients across the tympanic membrane impair hearing sensitivity [15] and speech intelligibility has been looked at as a function of altitude in aircraft [16]. However, the long and slow pressure gradients in airplanes provide opportunity to equilibrate middle ear pressure and induce hypobaric hypoxia with increasing altitude. To our knowledge, no data is available to discern whether the small and rapid transient pressure variations occurring inside high-speed trains passing through tunnels can affect intelligibility of speech.

[16] Wagstaff AS, Tvete O, Ludvigsen B. Speech intelligibility in aircraft noise as a function of altitude. Aviation, space, and environmental medicine. 1999;70(11):1064-9. Epub 1999/12/23. PubMed PMID: 10608602.

Question 2: In addition, although the direction of the speaker that reproduces noise in the experimental environment is toward the subject, it is considered that a sufficient diffused sound field will be realized only when the direction of the speaker faces the wall. If not, the speaker setup used by the author needs to be compensated for because of the directional influence of the sound source.

Figure 1 in the manuscript represents only a schematic of the experimental assembly and the depicted orientation of loudspeakers serves illustrative purposes; it is different from the actual orientation of the loudspeakers in the test facility. The pressure chamber used in this study is a former deep diving facility; it is essentially a horizontally oriented metal cylinder (see Photo P1 and P2 below). This barrel-shape of the facility is well suited to create homogeneous noise environments, which we have already done in previous studies simulating train and airplane environments (e.g. Sanok et al. 2015). We used four loudspeakers for playback of the background noise, two at both ends of the cylinder, each facing the round chamber wall. In addition, two separate loudspeakers were used for announcement of the test words. These two were installed face-up at the top “ceiling” of the round cylinder. Each loudspeaker was controlled by an individual DSP, providing complete control over the sound environment inside the chamber. Prior to the experiment the sound at each seat was specifically fine-tuned, using a calibration microphone at ear height. Due to this established procedure we have confidence in the quality of the acoustic environment realised during the experiment. However, even if acoustic differences between seats would have existed, this would not invalidate the experimental results. Due to the within-subject design and the fact that participants did not change seats during the study procedure the intra-individual differences between the pressure-change and no-pressure-change condition would still be a valid representation of the effect of this variable.

We added detailing sentences to Material and Methods:

The pressure chamber used for the experiment is a former deep diving facility, essentially a horizontally oriented metal cylinder, capable of producing rapid and highly accurate pressure variations [11]. The barrel-shape of the facility is well suited to create homogeneous noise environments and has been used for this purpose in previous studies simulating train environments [20]. We used four loudspeakers for playback of the background noise, two at both ends of the cylinder, each facing the round chamber wall. In addition, two separate loudspeakers, installed at the round ceiling, were used for announcement of the test words. The sound at each seat was calibrated using a microphone at ear height.

and Discussion:

We studied four participants during each test session and used an established calibration procedure to assure the quality of the acoustic environment [20]. But since the calibration was performed without having participants, i.e. four human bodies, inside the chamber, it is still possible that the acoustic properties at the different seats may have varied slightly during the actual experiments. However, the within-subject design, in combination with the fact that participants did not change seats during the procedure, ensures that the intra-individual differences in words understood are an unbiased representation of the effect of pressure changes on speech intelligibility.

[20] Sanok S, Mendolia F, Wittkowski M, Rooney D, Putzke M, Aeschbach D. Passenger comfort on high-speed trains: effect of tunnel noise on the subjective assessment of pressure variations. Ergonomics. 2015;58(6):1022-31. Epub 2015/01/20. doi: 10.1080/00140139.2014.997805. PubMed PMID: 25597694.

Photo P1: interior view of pressure chamber Photo P2: exterior view of pressure chamber

We hope our explanations show that we have paid great attention to every detail of our experiment, in order to make the design as well as the implementation of the study as scientifically sound as possible. We therefore respectfully disagree with the notion that, due to a lack of rigour in our experimental procedure, our manuscript was not suitable for publication in PLOS ONE.

Answers to comments from Reviewer 2

The paper report a study about intelligibility in trains due to changes in pressure, performed by means of measurements and questionnaires. The study is correctly performed, but no significant results are obtained. This will not be a paper that can leave the mark, but is still publishable.

Minor correction are reported to the authors:

Question 1: I suggest to avoid the abstract style used, which is too analytic. It is important to report what are the background, objective, methods, and so on, but it is not important to read them explicitly written.

We removed all structural-words from the abstract to make it more readable.

Please see Abstract in the revised version of the manuscript.

Question 2: A result of 0.7 plus or less 4.5 is a very bad number in math. It is simply very bad to read and should be reported in some different ways.

We agree, it may feel funny to read of 0.7 ± 4.5 words, since 0.7 words do not make sense in the real world. And since we use the Wilcoxon signed rank test we thought initially of reporting the intra-individual differences by median, as measure of central tendency, and quartiles to describe variability. However, Proudfoot et al. (2018) showed that describing paired count data by mean and standard deviation is a valid approach and this parametric representation has the advantage that it speaks naturally to the “internal statistics processor” of most people, i.e. the interpretation of mean = 0.7 words ± 4.5 SD as a non-significant result will be self-evident for most people, while a non-parametric representation of the data would not be. We are reporting a negative result of a sufficiently powered study, unfortunately still not a “familiar thing” for many researchers, and think it is therefore important for our results to be unambiguous and easy to interpret. For this reason we would like to keep the valid but odd-looking parametric representation we chose in the initial manuscript. However, we reworded all passages describing these numbers to improve readability and to make the underlying arithmetics very clear:

In Abstract:

On average, a participant understood 0.7 fewer words following a pressure change event compared to listening in steady ambient pressure. However, this intra-individual differences varied strongly between participants, standard deviation (SD) ± 4.5 words, resulting in a negligible effect size of 0.1 and the Wilcoxon signed rank test (Z=-1.26; p=0.21) did not distinguish it from chance. When comparing decreasing and increasing pressure events an average of 0.2 fewer words were understood (± 3.9 SD).

In Results:

When comparing the number of correctly understood test words announced immediately following a pressure change event to the number correctly understood in steady ambient pressure within each participant, the average understanding was 0.7 fewer words in the former condition. Due to the high variability of this intra-individual differences between the study participants, it has a standard deviation (SD) of ± 4.5 words, this equates only to a negligible effect size of 0.1 and the Wilcoxon signed rank test (Z=-1.26; p=0.21) does not detect it to be distinguishable from chance level. When comparing decreasing and increasing pressure events within each participant the average understanding differed by 0.2 words (± 3.9 SD), but this was not formally tested due to our hierarchically ordered hypothesis and lack of statistical significance of the primary hypothesis.

[22] Proudfoot JA, Lin T, Wang B, Tu XM. Tests for paired count outcomes. General psychiatry. 2018;31(1):e100004. Epub 2018/12/26. doi: 10.1136/gpsych-2018-100004. PubMed PMID: 30582120; PubMed Central PMCID: PMCPMC6211281.

Question 3: Conclusions are a bit short and should summarize better the paper. Furthermore, I suggest, and kindly ask, to mention at the end a sentence giving attention to the very important topic of railway noise impact over citizens, which is still very annoying and deserving attention and prevention. “However, railway noise still remains an issue for people living in the nearby of railway axis, deserving community attention and careful planning in order to avoid citizens’ complaints and annoyance [Licitra, G., Fredianelli, L., Petri, D., & Vigotti, M. A. (2016). Annoyance evaluation due to overall railway noise and vibration in Pisa urban areas. Science of the total environment, 568, 1315-1325; Bunn, F., & Zannin, P. H. T. (2016). Assessment of railway noise in an urban setting. Applied Acoustics, 104, 16-23.].

We extended the conclusions section in the manuscript to provide a more complete summary of our findings. We also emphasised that our experiment only addressed the isolated effect of pressure changes inside of trains and point to the general issue of railway noise:

The present study found that fast transient pressure changes in itself did not interfere with speech intelligibility and therefore do not appear to be a risk factor for disruption of verbal communication. This suggests that driver assistance systems can safely make use of speech to communicate relevant information. However, speech intelligibility requires sufficient SNR and a limitation of the overall sound pressure level. The train driver must therefore operate in a sufficiently quiet environment. While this may be achieved by acoustic-insulation of the train body, it is highly desirable to aim for an overall reduction of noise emission by the train, since railway noise is not only a relevant issue inside the cabin, but also an overarching problem for residents in proximity of railway lines [28], causing annoyance and sleep disturbance [29].

[28] Licitra G, Fredianelli L, Petri D, Vigotti MA. Annoyance evaluation due to overall railway noise and vibration in Pisa urban areas. The Science of the total environment. 2016;568:1315-25. Epub 2016/01/19. doi: 10.1016/j.scitotenv.2015.11.071. PubMed PMID: 26775834.

[29] Pennig S, Quehl J, Mueller U, Rolny V, Maass H, Basner M, et al. Annoyance and self-reported sleep disturbance due to night-time railway noise examined in the field. The Journal of the Acoustical Society of America. 2012;132(5):3109-17. Epub 2012/11/14. doi: 10.1121/1.4757732. PubMed PMID: 23145596.

Submitted filename: Answer-Letter_PONE.docx

Click here for additional data file.

7 Apr 2020

Influence of transient pressure changes on speech intelligibility: implications for next-generation train travel

PONE-D-19-26567R1

Dear Dr. Rooney,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Qiang Zeng, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This study is to investigate the effect of transient change of atmospheric pressure on speech intelligibility in travel train. It seems to be a good complement to the two comments previously pointed out. The reason for selecting SNR as 0 dB and the explanation for the experimental setup seems to be well complemented.

In particular, the originality of this study is very high because studies on the effect of environmental factors such as ambient pressure on speech intelligibility have been tried very little, so it is considered that this paper is suitable for publication in PLOS ONE.

Reviewer #2: The authors followed all my suggestions, and my comment is now positive.

The paper is now ready for being published

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

9 Apr 2020

PONE-D-19-26567R1

Influence of transient pressure changes on speech intelligibility: implications for next-generation train travel

Dear Dr. Rooney:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Qiang Zeng

Academic Editor

PLOS ONE