Reversing Goes Broadband

Multi-frequency reversing alarms help improve safety and reduce noise pollution

By Prof. D.J. Withington, School of Biomedical Sciences, University of Leeds

When designed correctly, auditory warning signals can improve operator performance and reduce accidents¹, while inadequate signal design can cause problems. In environments in which there is a serious mismatch between the perceived (psychoacoustic) urgency of a warning and its situational urgency (the urgency associated with the state or condition that the signal represents), the listener may not perceive the urgency of the situation that the signal is trying to communicate¹. When auditory warnings are not detectable within environments such as noisy workplaces, accidents may happen because a warning signal is either not heard² or is heard but not heeded, such as when the signal is sounded repetitively and ceases to convey a sense of danger or urgency. In other instances auditory warnings are so loud and distracting that people deactivate them, as with highly repetitive auditory warnings on rapid-cycle machines such as reversing alarms on industrial vehicles³.

Reversing safety should be addressed seriously by a range of groups including vehicle operators, policy-makers, researchers, health and safety specialists and driver trainers⁴. Eighty-six percent of reversing accidents occur at operators’ own premises or at collection/ delivery points⁴. A recent HSE report⁵ states that a major category of accident is the ‘slow-moving-vehicle reversing accident’, particularly involving fork-lift and other lift-truck vehicles, which represent a major traffic hazard in all countries. These conclusions were based on statistics from numerous sources and included the fact that every year there are about 8,000 lift-truck injury accidents in the UK alone (the figure rises to 68,000 in the US). Across all industries vehicle accidents account for 25% of workplace fatalities (the figure rising to 60% in the quarrying industry), nearly a quarter of all deaths involving vehicles at work occur while the vehicle is reversing, and from 1980 to 1990 there were 2,517 reportable accidents in the UK due to reversing heavy-goods vehicles. Having acknowledged the problem of reversing accidents, it is imperative that a vehicle’s reversing alarm should be well designed. Basically, there are two types of reversing alarm: conventional alarms (based on a single frequency) and broadband sound alarms (which use a multi-frequency range). This article compares the efficacy of the warning sound used by both types of alarm.

SCIENCE BEHIND REVERSING ALARMS

There are three factors that a reversing alarm should satisfy to perform optimally. These are: its ability to alert a listener, the locatability of the sound, and the impact the sound has on perceived noise nuisance.

Alerting sounds

An alerting sound is one with perceived urgency and detectability. Patterson6 proposed a warning signal design methodology. His design specified a brief pulse of sound as a basic ‘building block’ for the signal. A pulse is defined as a sound with an onset, an offset (decay time) and a specific duration, which is contained within one amplitude envelope. The pulse could be repeated several times, with intervals of silence between the pulses. The resultant unit, referred to as a burst of sound, forms the basis of a complete warning sound.

Both conventional (based on single tones) and broadband alarms are based around a simple principle for determining pulse duration and rate. SAE J994 states that the cycles of sound-pressure-level pulsations from the alarm shall be 0.8–2.2Hz. Both conventional and broadband alarms conform to this requirement. Additionally, the duration of the ‘on’ interval is equal to that of the ‘off’ interval (± 20%). Thus, both conventional and broadband alarms comply with the concept of warning signal design.

Work by Edworthy et al¹, and others^{7, 8, 9 & 10}, has found that many different acoustic parameters can be used to produce measurable differences in perceived urgency. The importance of creating a sense of urgency in a warning is twofold. First, to achieve the best mapping between the situation and the warning, a level of perceived urgency in the warning sound appropriate to the level of urgency of the situation is desirable. Secondly, in situations where a number of warnings may sound in quick succession, the listener will need to prioritize his/her responses. Perceived urgency defines how alerting, insistent and attention-grabbing a particular sound is.

The main sound parameters that create urgency are predominant frequency, rhythm, repetition speed and frequency range¹. A higher frequency increases urgency and a regular 1s ‘on’, 1s ‘off’ rhythm is perceived to be both effective and acceptable, while a faster rate creates greater urgency but tends to become insistent and irritating. Finally, a broad frequency range is the most urgent.

Assessing conventional and broadband alarms with regard to these parameters leads to the conclusion that both types of alarm (based on pulse duration and repetition rate) are effective alerters. The difference between broadband and conventional alarms, however, lies in the frequency content. Conventional alarms have a virtually nil frequency range whereas broadband alarms, by definition, have a multi-frequency range. This is their unique characteristic and it provides unique benefits as a result.

Locatable sounds

The ability to locate a sound source is an evolutionary prerequisite for survival. For example, the crack of a twig, which may signal the approach of a predator, demands instant reaction as soon as the audible signal is perceived. Similarly, a rustle of leaves may indicate to a predator where its prey is hiding and the catching of its food will depend upon locating that position. Given the right type of sound, a human can locate it to an accuracy of about five degrees¹¹. This level of accuracy is less than that for visual spatial acuity but more than adequate for survival purposes.

Brain location
One particular part of the central nervous system plays a vital role in the detection of and, equally importantly, response to a sound source. This area is part of the mid-brain and is known as the superior colliculus (SC)¹². Neurophysiologists studying the properties of neurones in the SC together with psychoacousticians studying human responses to sound have developed an understanding of how the brain processes information relating to a sound source and, importantly, what type of sound is needed for a degree of accuracy to be achieved. It has long been recognized that locating a sound source requires a vast amount of neural processing¹³. Only certain types of sounds are inherently locatable and what is crucial is that they contain a large frequency spectrum, ie broadband sound. Single-frequency tones or combinations thereof, or narrowband sounds, cannot be located. To understand why it is necessary to consider the cues given by sounds which are recognized by the brain.

Location cues
Humans can hear a frequency range from approximately 20Hz to 20kHz, although this range diminishes with age. Three main clues allow the brain to locate sound. The first two are known as binaural cues because they rely on the fact that a person has two ears, separated by the width of his/her head. A sound emanating from either side of the mid-line arrives first at, and is loudest at, the ear closest to it. At low frequencies the brain recognizes differences in the time of arrival of the sound between the ears (ITD), and at higher frequencies the salient cue is the loudness/ intensity difference between the sound at each ear (IID). The use of these two cues is known as the ‘duplex’ theory and was proposed by Lord Raleigh as long ago as 1877 (fig. 1).

For single frequencies, however, these cues are spatially ambiguous. The inherent ambiguity has been described as the ‘cone of confusion’. This is because for any given frequency there are numerous spatial positions with identical timing/ intensity differences; these can be represented graphically in the form of a cone whose apex is at the level of the external ear. The cone of confusion is the main cause of human inability able to locate pure tones^{14 & 15}.

The third clue to sound location is the head-related transfer function (HRTF)16. The HRTF refers to the effect the external part of the ear (pinna) has on sound. As a result of passing over the bumps or convolutions of the pinna, the sound is modified so that some frequencies are attenuated and others are amplified (fig. 2). Although there are certain generalities in the way the sound is modified by the pinna, the HRTF of any one person is unique to that individual. The role of the HRTF is particularly important in determining whether a sound is in front or behind. In this instance, the timing and intensity differences are negligible and there is consequently very little information available to the central nervous system on which to base a decision of ‘in front’ or ‘behind’. Therefore, to locate a sound source and overcome the ambiguities inherent to single tones, a multi-frequency band is essential and the broader the better.

The US Standard for reversing alarms, SAE J994, sets a frequency range for conventional single-frequency alarms at 700–2,800Hz. Most speaker-based alarms emit a single frequency in a range from 900–1,700Hz and the much cheaper piezo-based models emit a single frequency in the range 2,500–3,500Hz.

Conventional alarms fall well short of a sound that will allow the listener to pinpoint it. Essentially, three regions of the human hearing range are utilized to locate a sound source: 1,000Hz and below, 3,000Hz and above, and 5,000Hz and above (see fig. 3). All three regions are required (simultaneously) for sound location. Conventional single-frequency alarms generally operate in the range that gives no location cues (1,000–3,000Hz). Even a 700Hz sound alone (falling within the first required region), withou the benefit of higher frequencies, will not allow the listener to locate its source.

By contrast, broadband sound covers the majority of the human hearing range. This sound is ideal for sound location and allows the listener to pinpoint a sound source immediately.

Noise annoyance

Noise pollution is now recognized as a major health hazard to which conventional reversing alarms are a major contributor. Now a worldwide problem, reports of noise pollution due to their proliferation come from as far afield as the US and New Zealand. People, including workers on site and even plant operators themselves, are often driven to distraction by the irritating bleep, bleep, bleep of reversing alarms echoing round the site. On the other hand, it is essential to give audible warning of a reversing vehicle to persons in the hazard area behind it. Communities near to quarries, construction sites, commercial premises, warehouses, distribution centres, loading and receiving bays, bus depots and wherever garbage is collected — which means just about every domestic premises — all have to suffer it. Some local authorities are threatening to prohibit their use, arguing that the noise-pollution problem overrides even safety considerations. Clearly something has to be done.

High alarm signal levels cannot compensate for poor warning design, and most contemporary researchers acknowledge the fact that sound levels that are too high can be disruptive in an emergency. The intensity of an alarm must allow the sound to be detectable above the background noise and must grab the attention of the listener without causing a startle reaction.

It is apparent from field measurements that a broadband sound alarm is very effective at identifying the location of the reversing vehicle. Additionally, there is a significant reduction in extraneous sound beyond the immediate area (ie directly behind the reversing vehicle). A broadband reversing alarm is equally effective at alerting the listener at around 5dB(A) less than a conventional reversing alarm and, additionally, is little heard either in front of or to the sides of the vehicle.

The following are some of the physical and psychoacoustic principles that explain these effects.

Equal SPL measurements and spectral analysis

A reading of sound pressure on an SPL meter (as per ANSI S1.4 (or IEC 60651) — specification for sound level meters) will ‘average’ the sound pressure in each frequency band and present a consolidated single-figure output, weighted as per the settings on the meter.

It is conventional to measure sound using the ‘A’ weighting (dB(A)) as this is generally accepted to be the closest representation of the actual response of the human ear. This ‘A’ weighting adds or subtracts a number of decibels from the SPL reading from each of the frequency bands in order to simulate the non-linear output vs frequency response of the ear.

The graph in figure 4 shows the outputs which could be expected from a conventional reversing alarm, when centered on 1,250Hz (black line), and a broadband reversing alarm (red line). The frequency range of the broadband is much larger than the conventional reversing alarm but is consistently at a lower level. These outputs could be obtained from a sound meter (and filter set), as per ANSI S1.4 & S1.11 (or IEC 60651 & 61260), set to the one-third octave range.

Although the broadband sound shows a lower sound-pressure level in each one-third octave band, the combined effect of these (when summed) is a level equal to the conventional reversing alarm — 100dB(A) at 1m.

The summation of decibels is as follows:

where spl1, spl2, spl3 etc are the individual one-third octave band sound-pressure levels.

The simulated ‘background noise level’ is also shown on the graph (dotted line). This is unrealistically flat but again demonstrates that the ‘flat level’ of about 52dB(A) in each one-third octave band can produce an overall SPL of about 66dB(A). This background noise level would be regarded as quiet for a quarry or construction site.

Sound vs distance

In a free field (open, 3D spherical space) sound dissipates from a point source according to the inverse square law and the reduction in dB is calculated as:

where r is the distance of the listener from the source. This results in the well known 6dB drop for each doubling in distance from the source. However, most sound sources are not ‘ideal point sources’ and hence have less than ideal sound distribution in all directions.

Therefore, ignoring any other environmental factors, the SPL will be reduced by about 36dB in each one-third octave band for a listener 60m away from the source. From the graph in figure 4 it can be seen that broadband levels will have dropped almost to background noise levels, whereas the conventional reversing alarm will still be some 10dB(A) above background.

In addition, various features absorb sound at different rates depending upon frequency content. The air absorbs sound faster (ie more per doubling of distance) in the higher frequency ranges. Atmospheric conditions (humidity, temperature, wind direction etc) also affect the rate at which sound is absorbed. There will be additional absorption by physical structures (buildings, fences, trees etc), which all have frequency-dependent absorption rates.

Finally, the psychoacoustics of sound influence the perception of which sounds are regarded as ‘pleasant’ or ‘intrusive’. People are more sensitive to sound in the region of 1kHz to 4kHz and perceive sound in this region to be louder than equally measured sound in other frequency bands (this forms the basis for the ‘A’ weighting system). People are also very sensitive to discrete frequencies (pure tones) and can detect the presence of these in quite high background noise levels.

Loudness is the most highly rated factor associated with annoyance for reversing alarms. Peak loudness, as typified by conventional alarms, has been identified in a US Department of Transportation study as the primary reason people describe a reversing sound as ‘annoying’17. The spread of sound pressure over a broadband range results in a ‘less-annoying’ sound.

SITE SURVEY DATA

A survey involving 1,477 vehicles fitted with reversing alarms, of which 313 were new broadband sound alarms, was undertaken recently. A range of companies participated in the survey and a questionnaire was directed at both site managers and site workers to assess their views on conventional and broadband alarms (described as the ‘new sound’). Questions 1–4 were asked about each type of alarm and questions 5–6 looked at the perceived differences between the two alarms.

• When you first heard the conventional alarm/‘new sound’ did you realize it was a warning alarm? Yes /No
• How long did it take you to associate the sound with a vehicle reversing? Immediately/ A few minutes/A few hours/Days/ Never
• When multiple vehicles are in use, is it important to identify which vehicle is reversing towards you? Yes/No
• When you hear the sound can you tell which vehicle it belongs to? Always/Mostly/ Sometimes/Rarely/Never
• Do you find the ‘new sound’ more or less irritating than the conventional/normal reversing alarm? Much less irritating/A little less irritating/Similar/A little more irritating/Much more irritating
• If there were people in the near vicinity of the operating vehicles do you think they would be more or less irritated by the ‘new sound’ compared to the conventional/normal reversing tones? Less irritated/ No difference/More irritated

The first two questions looked at the alerting nature of the alarm, questions 3 and 4 asked about safety issues, and 5 and 6 focused on the issue of noise annoyance.

Results

In response to question 1 both the conventional and broadband sounds were rated at 100%, ie both were considered to be warning alarms.

The results from question 2 are shown in figure 5.

Ninety percent of participants stated that they thought it was important to be able to tell which vehicle was reversing towards them (question 3). The accuracy of the conventional and broadband sounds in identifying a vehicle (question 4) is shown in figure 6.

The responses to questions 5 and 6 are shown in figure 7.

Discussion

The questionnaire data revealed information on the three crucial aspects required in an optimal reversing alarm. First, the alerting nature of the sound, as addressed by questions 1 and 2. Both the conventional and broadband sounds are recognized as a warning alarm. The data in response to the second question about how long it took to associate the sound with a vehicle reversing, also show that both alarms are quickly associated with the event they are warning about. The data on the conventional alarm are probably skewed in favour of a quick response due to the fact that such alarms have been in use for over 20 years and it is unlikely that respondents could really remember how long it took them to associate the conventional sound with the vehicle reversing.

From a safety point of view, it is not surprising that the almost every operator believes it is important to identify which vehicle is actually reversing (question 3). The conventional alarm data of listeners’ ability to distinguish which vehicle is reversing — ‘mostly’, ‘sometimes’ or ‘rarely’ — indicates a serious safety issue. In contrast, when questioned about the broadband reversing alarm the ‘always’ response was 80% (question 4, figure 6). These field data confirm the theory of what makes a locatable sound.

Finally, the noise annoyance issue was examined (questions 5 and 6). All respondents stated that the broadband sound was less annoying than the conventional alarm even though some resistance to a new sound might have been expected. The HSE and other safety bodies (eg OSHA) are aware that site personnel frequently deactivate an alarm that is irritating, the deactivation often being ignored until the vehicle is examined following an accident. Reports of sabotaged alarms range from cut wires to muffled speakers and even a well-aimed sledgehammer. No reports have yet been received of the sabotage of a broadband reversing alarm.

SUMMARY

Theoretically, both broadband and conventional alarms fulfil the requirements of an alerting sound. However, the conventional alarm suffers from two major drawbacks: its non-locatability by the listener and its noise nuisance. Noise pollution has surfaced recently as a major issue worldwide and the broadband reversing alarm solves this problem at a stroke.

Site questionnaire data for both types of reversing alarm confirm the scientifically predicted outcome that conventional alarms are potentially dangerous. Although alerted, workers on multi-vehicle sites can be totally confused as to which vehicle is reversing. The broadband alarm, on the other hand, provides instant identification of a reversing vehicle. In addition, the elimination of noise pollution provides an immeasurable benefit to site workers and local communities alike. As a result, broadband sound alarms represent a significant step forward in workplace safety and make a major contribution to a less noisy world. 

REFERENCES

 

1. EDWORTHY, J., LOXLEY, S., and I. DENNIS: ‘Improving auditory warning design: Relationships between warning sound parameters and perceived urgency’, Human Factors, 1991, 33, pp 205–31.
2. WILKINS, P.A., and W.I. ACTON: ‘Noise and Accidents – a review’, Annals of Occupational Hygiene, 1982, 2, pp 249–60.
3. HAAS, E.C., and J. EDWORTHY: ‘The perceived urgency and detection time of multitone auditory signals’, in Human Factors in Auditory Warnings, Eds N.A. Stanton and J. Edworthy, 1998, Ashgate, England.
4. MURRAY, W., MILLS, J., and P. MOORE: ‘Reversing accidents in UK transport fleets 1996–97’, University of Huddersfield Report, 1998.
5. Health and Safety Executive, Report Number 358, ‘Improving the safety of workers in the vicinity of mobile plant’, 2001.
6. PATTERSON, R.D.: ‘Guidelines for Auditory Warning Systems on Civil Aircraft’, CAA paper 82017, London: Civil Aviation Authority, 1982.
7. HELLIER, E., and J. EDWORTHY: ‘Quantifying the perceived urgency of auditory warnings’, Canadian Acoustics, 1989, 17, pp 3–11.
8. EDWORTHY, J., and N.A. STANTON: ‘A user-centred approach to the design and evaluation of auditory warning signals: 1 Methodology’, Ergonomics, 1995, 38, pp 2,262–80.
9. HAAS, E.C., and J.G. CASALI: ‘Perceived urgency of and response time to multi-tone and frequency modulated warning signals in broadband noise’, Ergonomics, 1995, 38, pp 2,313–26.
10. HAAS, E.C., and J. EDWORTHY: ‘Designing urgency into auditory warnings using pitch, speed and loudness', Computing and Control Engineering Journal, 1996, 7(4), pp 193–8.
11. MAKOUS, J.C., and J.C. MIDDLEBROOKS: ‘Two-dimensional sound localization by human listeners, Journal of the Acoustical Society of America, 1990, 87, pp 2,188–200.
12. STEIN, B.E., and M.A. MEREDITH: The Merging of the Senses, Cambridge, MA: MIT Press, 1993.
13. KNUDSEN, E.I., DU LAC, S., and S.D. ESTERLEY: ‘Computational maps in the brain’, Annual Review of Neuroscience, 1987, 10, pp 41–65.
14. BLAUERT, J.: Spatial Hearing, Cambridge, MA: MIT Press, 1997.
15. WIGHTMAN, F.L., and D.J. KISTLER: ‘Sound Localization’, in Human Psychophysics, Eds W.A. Yost, A.N. Popper and R.R. Fay, 1993, New York: Springer-Verlag.
16. CARLILE, S., and A.J. KING: ‘Auditory Neuroscience: from outer ear to virtual space’, Current Biology, 1993, 3, pp 446–8.
17. HARPSTER, J., HUEY, R.W., LERNER, N.D., and G.V. STEINBERG: ‘Backup warning signals: driver perception and response’, US Department of Transportation DOT HS 808 536, 1996.
18. STACH, B.A.: Comprehensive Dictionary of Audiology, 1997, Williams & Wilkins, Maryland.