Published Sept 2012

and why current regulations have limited effectiveness

Environmental noise in urban and suburban areas arises from many sources. The unwanted noise levels in our larger cities and suburbs have been ever-increasing. By far the prevalent cause of unwanted noise is due to transportation. Think air, rail, subway and road traffic. Other urban noise sources such as from building and infrastructure construction tends to be transitory at any one location. In the suburbs, the rasp of lawn-mowers and weed-whackers is inevitable, as are noisy parties, but these are usually of relatively short duration. This is not to say these noise sources are blameless. However, noise from road traffic is in need of considerable attention because it is entirely pervasive, occurs practically around the clock and affects everybody. Not everyone is unfortunate to be living under aircraft approach paths, near airports or railway lines, but virtually everybody is adjacent to a road.

Unwanted noise has wide-ranging effects on human health to a far greater extent than merely to cause damage to hearing. Noise results in cardiovascular issues, sleep disturbance, headaches, nausea, impaired task functioning, depression and is implicated in many other health problems as well as inducing anti-social behaviours. (references 3,4,12,13,17,23) These unwanted health problems occur with noise levels which are a lot less loud than that which would cause hearing damage. Effects are also cumulative. The costs to health-care systems is enormous and researchers are finding more and more stress related problems that can be attributed to noise.

Governments, city planners and local councils are, in general, aware of the noise problem, and try to provide urban planning guidelines and rules to mitigate noise, but prefer to take 'baby-steps' and are over-cautious in order to avoid the creation of higher compliance costs. In most cases, authorities do not apply sufficient urgency or big enough solutions to the problem. There are already a large number of studies and comprehensive research documents published on the subject. No more are needed. Instead, some definitive progress is called for, which will certainly have a cost. However, studies have already shown that the costs of healthcare triggered by unwanted noise is more than a hundred times than what is currently spent on reducing or mitigating noise from vehicular traffic.

In Australia for example, recent findings released by the Australian Bureau of Statistics show that at least a third of the population rate noisy vehicles as their leading social concern. (ref. 9) A study from the N.Z general social survey 2008 showed that that the most common neighbourhood problem cited was that of unwanted noise. Inhabitants of a civilised society with enlightened governance should be able to live and work in a quiet environment. Even ancient Rome had rules against horse and cart clattering along cobbled streets at night.


Nearly all road vehicle types can create excessive noise. Partly, this is a matter of inadequate attention to low noise design and partly due to inconsiderate driving manner.

Trucks, large and small, petrol or diesel.

Who hasn't been affronted by large diesel trucks grinding away from traffic lights, red-lining the engine through every gear. Not to mention enduring the massive pulses of low frequency sound when they use the engine or exhaust brakes to slow down. Many trucks of all sizes are rattle-traps particularly when they pass over uneven road surfaces.


Ageing fleets of city buses are a noise problem. With relatively large engine capacity, when these diesels lurch away from their stops in low gear, they produce very loud engine sounds. The noise includes low frequency pulses that can be felt as a vibration in one's chest when walking close by. The sudden release of air from the pneumatic brakes just prior to a bus departing a bus-stop is heart-attack inducing for all bystanders as well. Here in Wellington, New Zealand, we are (arguably) blessed through still having about one quarter of the bus fleet as electric trolley buses. The engine noise from these comprises mainly a whine from the motor as they accelerate. I say arguably since trolley buses have their own set of problems such as high capital cost, high maintenance costs and with the need for arrays of unsightly overhead wires which will be a liability in any civil emergency.

Courier and trades vans.

From observation, the noise issues from this sector have a lot to do with the manner of driving. For reasons which are unclear, courier and trades van drivers appear to need to spend as little time as possible between the stopped state and the speed limit. They accelerate up through the gears as rapidly as possible and also brake aggressively, often using engine braking in low gears. Many drivers appear to need to operate loud music with windows open, thus exacerbating the net noise problem.

Cars, especially older models and those with after-market exhaust systems fitted.

Because cars comprise at least 85 percent of road traffic, they are also the greater problem. New Zealand and most similar societies have regulations around environmental noise levels from vehicles. There is no longer any car manufacturing in this country and so all makes and models of cars imported have to meet noise type approvals. The regulations here are based more or less on those in the jurisdictions where cars are manufactured, it is assumed if they meet noise approvals there, they will do so here. Western societies such as the EU, USA, Australia and here do recognise the need to reduce noise emissions from cars and from time to time, they reduce the noise limits that new cars must conform to. However, this is a slow process. The EU has recently introduced lower noise limits for new motor vehicles after recognising that the previous amendments 20 years ago had made a negligible difference to statistical noise levels from road traffic. The New Zealand regulations were amended last in 2009 but the limits are loosely based on EU regulations prior to their previous update. Of course, to avoid owners of older cars having to 'upgrade' to the new standards, the noise allowed is higher for older vehicles. Because of this, any statistical reduction of vehicle noise takes a long time to occur, as it depends on how often owners replace their cars. It has been noted that we in N.Z tend to keep our cars for longer than do people from other comparable societies and so any improvements to noise levels will occur only in the longer term.

There is a group of people who like to modify their cars and in doing so, make them louder. I see no sub-group amongst the 'car enthusiast' ilk which modify their cars to make them quieter. Many, although not all, of these 'car enthusiasts' drive in an anti-social fashion. They have not yet overcome the juvenile 'look-at-me' visceral impulses which see them drive their cars aggressively around the streets having fitted after-market exhaust systems that are designed to be noticed. N.Z regulations attempt to capture the modified exhaust set, by including clauses that say an exhaust may not be louder than the stock exhaust, unless it actually does still comply with noise limits.

I see no sub-group amongst the 'car enthusiast' ilk which modify their cars to make them quieter
However, compliance has limited effectiveness since policing rates are low. In 2003, only 1000 noise abatement notices were issued annually across the country. This represented 0.1% of noisy cars back then as adjudged by roadside testing. Having limited resources, police have more urgent priorities. Furthermore, even stock exhausts on some cars produce sound levels which are annoying, more especially when such cars are driven in an aggressive style.


That which has been said of cars is largely true for motorbikes. There are quiet motorbikes and noisy ones. There are anti-social bike riders who like to create a lot of noise. All else being equal, it is harder to muffle a bike than it is for larger vehicles. There is not much space to include an effective muffler. Current regulations build in this assumption and allow higher noise limits for bikes. I regard the argument as flawed since I have witnessed countless times, large bikes that just purr past making very tolerable noise, and also bikes that create sonic-boom-like rackets that would definitely exceed health and safety levels. If low noise can be achieved, then make it a regulation to do so. Even Harley-Davidson is taking noise seriously and are embarking on a generation of bikes including a design goal of specifically reducing noise.

Emergency vehicles.

It may seem indecorous to include this vehicle class, but there is no doubt that emergency sirens are noise pollution from a road vehicle. Admittedly there is no remedy to this except to say that the intrusiveness is infrequent and residents are tolerant. Emergency operators are perceptive enough to restrict use of sirens where they can do so. Perhaps over time, if general vehicle noise is lower or when we all have electric cars, the sirens can also be turned down.


For a mobile vehicle at lower speeds, exhaust noise is usually regarded as dominant. However at speeds between 50km/hour and 70km/hour, road noise takes over, coming mainly from tyre/road contact. The type of road surface plays a big role in this. The ubiquitous coarse-chip seal on N.Z roads is noisy. Authorities do apply quieter fine and medium chip seals to roads in residential areas of higher traffic density. Research is on-going to find quieter road surfaces, which are durable and cost-effective.

Gearbox and transmission noise is significant mainly from heavier vehicles such as trucks, buses and sometimes vans.

For noise sufferers on city and suburban streets, exhaust and engine noise will be more intrusive, while those who live on arterial routes or near motorways also have to cope with running and road surface contact noise generated by vehicles.

The 'thud-thud' noise heard from high power car stereos will be more invasive to residents on urban and suburban streets than say on arterials and motorways, where the annoyance passes quickly, however that does depend on the numbers of soon-to-be-deaf drivers that are passing.


In New Zealand, we have various organisations who have interests in environmental noise.

Health and safety at work

Our health and safety organisation (OSH) within the department of labour is tasked with ensuring noise exposure levels in the workplace do not exceed guidelines, which are set to avoid hearing damage. OSH measure 'noise exposure' levels, which are a time averaged figure, usually over an 8 hour period. The current limits are:

Noise exposure: L Aeq(8hr) =85dB; no peak to exceed 140dB (unweighted).

What this means is that a worker cannot be subjected to a sound level of more than 85dBA continuously (based on 8 hours). However the sound levels can be higher over shorter periods: (to get the same exposure result) e.g 88dBA for 4 hours, or 91dBA for 2 hours are both equivalent. If a 16 hours shift was worked, the limit becomes 82dBA continuous noise. To capture varying noise levels, an averaging meter (dosimeter) is used which continuously measures and calculates the accumulated noise exposure levels.

Recently, it has become recognised that those levels are too high to prevent hearing loss and a 5dB lower limit is proposed which is already in use in some countries. The use of A-weighting is a little controversial for the purpose of measuring sound levels associated with hearing damage. More on weighting later in the article.


City and regional councils commission city planning and city councils also enforce noise complaints from residents. Such noise is from construction, excess use of machinery, parties, dogs and so on. The councils here do take noise complaints seriously, however they have limited resources for enforcement and it is sometimes not easy to get long term cessation of noise created by our more determined residents. Councils do not enforce noise from road vehicles.

Resource and environmental planning

The Ministry of the Environment looks after resource management which encompasses matters like noise mitigation. This ministry, and other related departments, commission surveys on traffic noise. See Quality Planning:Land transport Noise.

Vehicle Noise

The N.Z Transport Agency (NZTA) determines certification requirements for vehicles, including noise emissions. The police have a role enforcing vehicle noise limits.

Link to NZTA vehicle noise rules-latest amendment.

NZTA rules say that all imported vehicles must comply with their noise rules which are based on the comprehensive ISO362 or equivalent standard, relevant to the class of vehicle. ISO362 is a drive-by test, which requires measurements of noise as a vehicle drives past the microphone at several combinations of acceleration, constant speed and in various gears. Needless to say the setting up of a venue to make this test is expensive and here in N.Z we have no established facility to replicate the drive-by measurements. Instead, NZTA allows for a static tailpipe noise test, which can be applied to routine in-service checks (WOF; COF), modified's, or built from scratch vehicles. This is based on ISO5130. These latter checks are referred to by LTSA as an 'objective noise test' or 'ONT' and can only be performed by licenced practitioners of the awkwardly named 'Low Volume Vehicle Technical Association' or LVVTA. There are only about 40 LVVTA noise inspectors in the entire country, and so owners who need or have been required to take their vehicle for an ONT may need to travel some distance.

The ISO362 drive-by test and the ISO5130 tailpipe test are not the same and produce different noise result figures, but NZTA have tables for each.

Not every vehicle needs a noise test at the time of renewing the warrant of fitness. Any vehicle which met the type certification when it first entered the country and hasn't been modified, automatically passes. This is sensible and places the burden of being tested for noise on to exceptions. The WOF inspector is charged with deciding whether a modified vehicle needs to go for an objective noise test. The inspector has to make the decision subjectively whether a vehicle is 'too loud'. He 'may' have available a simple noise meter, and be trained to use it. If so, he can elect to give a car a 'quick noise check' to assist the judgement of whether the vehicle is too loud.


A point by point discussion

The question is whether current measurement practices and regulations are sufficient to reduce the noise pollution from vehicles over a reasonably short time. New Zealand authorities tend to adopt a light-handed regulatory regime. This philosophy has the advantages of making compliance costs lower, or able to be spread over a long period, but mainly reduces costs of policing. Light, or soft regulation tends to place the burden of 'doing the right thing' on owners and commercial operators. The majority of people are willing to comply with sensible regulations, but there are commercial imperatives plus individuals who like to 'stretch' the rules. This means that an effective system of enforcement is needed in any case.

Time averaged noise exposure levels; L(A)eq, don't tell the full story

First, as mentioned earlier, OSH and the department of labour use time averaged exposure levels (LAeq) to set limits for the purpose of avoiding hearing damage to workers. There is no issue with using equivalent sound level for that purpose.

It is interesting that surveys of traffic noise rely solely on equivalent sound levels as well, even though surveys are not concerned about noise levels which would damage our hearing. Typical traffic surveys usually split up a day into 3 periods; day, evening and night; calculating an equivalent sound level for each period. This method means that a zone having 22,000 vehicles passing a day will correctly be regarded as noisier than a place with only 330 vehicles a day. So far, so good. The problems I see with the method are:

  1. The method does not capture loud peaks. If a few very noisy vehicles pass by, the calculated L(A)eq will not change.
  2. Results are given A-weighting. More on this later in the discussion.
  3. Results of such traffic surveys appear to be used for urban planning such as whether to screen residents and workers from traffic noise at any location, but do not appear to be shared and used for the purpose of setting vehicle noise limits. It would seem more useful to restrict the source of the noise as to mitigate its effects.

Regarding point 1, there appears to be no doubt that continuous traffic noise causes medical complications, but is not regarded as being responsible for ear damage and loss of hearing.

Impulsive noise; that is bursts of loud noises occurring infrequently within an otherwise quiet environment are also 'annoying' and also create medical issues, although the complications are liable to be different than those caused by continuous unwanted noise. It might be argued that a handful of noisy vehicles passing in an otherwise very quiet street are more annoying than having a continuous drone of traffic. Office air-conditioning is an example of where the steady whoosh and hum is used to 'mask' transient noises which would otherwise be very noticeable, resulting in workers suffering loss of concentration.

Can 'annoyance' due to noise be classified with a single number

This is one of the classical problems where we try and equate a subjective response (annoyance in this case) to an objective measurement. As well as trying to enumerate emotional responses like annoyance, we are also attempting to find a number (of decibels) below which, medical complications would not be caused. These are the essential reasons behind needing a noise limit for road vehicles.

Vehicle noise tests, whether by the ISO362 drive-by version, or the stationary tail-pipe method, produce a single number (of decibels) relative to the human threshold of hearing. The drive-by method is more inclusive; that is it creates the decibel number from the worst case of acceleration, steady speed and also using several gears. Not only that, the drive-by result includes all emitted sound; not just from the exhaust-pipe, but including noise such as from tyres, the engine bay and vehicle transmission.

The alternative, simpler test, based on ISO5130, used in New Zealand just measures the tailpipe noise. The decibel number is always greater than for the drive-by test, because the test microphone is only half a metre from the tailpipe. So, can the tailpipe test be representative of total vehicle noise in a real world situation?

The authors of 'Review of Methods Used For The Control Of Excessive Noise From New Zealand Road Vehicles' (reference 1), performed a drive-by test and an exhaust test on six vehicles in 2004. The result showed there is a reasonably consistent relationship between the two methods such that the exhaust test returned figures averaging 13.6dB higher than did the drive-by test, with a 95% confidence range of 1.2dB either side. The interesting point here is to note that if all the vehicle noise came from the tailpipe, then the difference between the two methods would be over 23dB, simply due to the differing measuring distance. Since the difference is much less, then it is clear that much of the total noise of a vehicle comes from other than the tailpipe. Their test vehicles comprised three small cars, two large cars and one large 4WD vehicle. No vans, trucks, buses or motorbikes were tested and it would be unwise to try and apply this same relationship to these other vehicle types.

Annoyance is an emotional reaction and has emotional origins. As mentioned above, most people are likely to be annoyed by a small number of loud vehicles on a quiet street which has little traffic. However, we are 'less annoyed' if that loud vehicle is, say, a recycling truck, than we are if the vehicle is an old bomb with a modified exhaust driven by a teenager racing up the street. This is true even if the noise levels were the same. Studies have shown that more irritation is caused by noise that is changing (accelerating car) than from a car at a steady speed, again, even when the acoustic noise level is the same. There is also the problem of the 'character' of noise. People also notice noise that stands out from the background, so if a machine has a steady whoosh or drone, plus a screech or whistle or a beat at lower volume, then the screech or repetitive beat would be more noticeable and annoying.

Next is a computer-generated example of two sounds which have exactly the same volume, but have a different character.
Play the short 6 second track.

The tone is an approximation of a 4-cylinder car driving by at a steady 4000rpm. The first 3 seconds might be equated to the car having a poor muffler, then for the last 3 seconds, it is given a different muffler. The second half (muffled part) eliminates all frequencies above 500Hz, leaving just the base cylinder and engine firing noise components. It would be surprising if you did not regard the first half as more annoying, but this is simply to illustrate that two sounds of the same volume can sound rather different.

The vehicle noise limits used in New Zealand

The current rules in New Zealand restrict the initial importing of vehicles to those that meet type testing according to drive-by testing methods per ISO362. This type-testing is not done in New Zealand. For in-service testing as well as for modified and scratch-built vehicles, an objective noise test based on a tailpipe measurement is available, if determined necessary by a vehicle testing officer or the police. These rules apply only to 'light' vehicles such as cars, vans, bikes and light trucks weighing under 3.5 tonnes. See references 19,20,21 for NZTA rules and guides. The link below is to the report to government supporting the 2009 amendment which allowed for a 5dB reduction in the tailpipe noise level to 90dBA applying only to post-1985 built cars entering service after June 2008. The intention of this amendment was to target the few percent of noisiest vehicles; mainly those with modified exhaust systems.
Link to 2009 amendment proposals.

Essential NZTA post-2009 target values:

TABLE 1: Importation Limits: (based on ISO362 drive-by methods)

TABLE 2: In-Service test limits as measured by a ISO5130 method tailpipe test:

     Car, vans, SUV's, people movers, small buses and light goods vehicles: Classes MA, MB, MC, MD1, MD2, and NA

Note that there are no tailpipe tests specified for buses and heavy goods vehicles as NZTA considers that these are not usually subject to modification.

Issues around the New Zealand methodology

The limiting values

Even if you accept that a single A-weighted decibel number is sufficient to describe vehicle loudness, and also annoyance, there is the question of whether these numbers are going to reduce overall traffic noise levels and eliminate the noise of worst-case offenders.

In Australia, authorities have a similar methodology of applying type-acceptance noise ratings measured by the current drive-by process to new vehicles in the first instance. Their dB noise value for cars is similar to N.Z although they have 3-5dB lower limits for higher powered vehicles; buses and trucks. Australia also allows for in-service tailpipe testing, but there, a signature measurement of a vehicle is taken at initial entry of that type into the fleet. A vehicle subsequently must not exceed that signature reading from the tailpipe test. Interestingly, from their published figures for relatively new vehicles that are not electric or hybrid, the tailpipe signature results range from 73dBA (Skoda Yeti) to 90dBA (Fiat 500), with an average of 80dBA. The results include small, medium and large cars from 2009 models. For hybrids, the tailpipe signature results range from 67dBA (Lexus CT200h hatch) to 77dBA (Prius C hatch), with an average of 72dBA.

New Zealand has just recently lowered their car tailpipe noise maximum to 90dBA, as outlined earlier for cars registered post-June 2008. It seems N.Z is allowing for worst case noise makers; a philosophy which will not have much effect on overall traffic noise levels and appears out of touch with what can be achieved. The extra leniency for motorbikes is due partly because of the proximity of the engine to the tailpipe, which adds to tailpipe noise during testing. I think there is plenty of room to reduce these limits in spite of this, since motorbikes can be made much quieter than what this test allows for at present and making allowance for the worst cases does not motivate manufacturers to improve things.

The EU community has just introduced new staged noise limits, although the full effect of these will not be seen until 2017. Limits for cars (on the drive-by tests) will become 68dBA for cars and up to 78dBA for trucks. Compare this with current N.Z importation rules based on drive-by testing of 81dBA and 88dBA respectively. Generally, it is reasonable to expect cars that meet new directives in the country of manufacture, to also easily meet N.Z regulations, although manufacturers are quite at liberty to build variants that are intended for 'less tough' countries, which would be cheaper to build through not having all the noise emission reduction features.

The regulations also allow for a modified exhaust system, as long as it does still meet the tailpipe noise limits in the schedules. This means that an owner could take an initially quiet car (one that easily met the noise criteria) and turn it into a car that 'just' meets the limits, thus legally contributing to an INCREASE in statistical traffic noise.

Reliance on subjective noise assessment for warrant of fitness

Currently it is up to WOF inspectors to subjectively assess whether a car is too loud. If they deem it so, they will fail the car and require the owner to take it for an objective noise test (ONT). In the first instance, any car seen to have an unmodified exhaust system automatically passes, unless the exhaust is faulty. I regard this process as too variable and too onerous for an inspector for three reasons:

Avoiding arguments.

Although the 'official' vehicle testing stations are more formalised, ordinary garages and dealerships will want to avoid failing vehicles for noise because it will soon become known that they are 'tough', and that will reduce their custom over time. Furthermore, inspectors are 'nice guys' and don't like failing vehicles, especially since it is a non-safety-related matter, because they are aware of the cost and inconvenience to the owner. Imagine the conflict of interest if the car owner was family, a friend, or friend of a friend, or neighbour. It is easier to tick the box and say the car was 'not noticeably louder'.


Vehicle testing facilities make money by testing as many vehicles as possible in a day. There is pressure on inspectors to get them done without undue delay. Again, it is more expedient to just tick the box than to have to take the car from the garage out into the yard, find a colleague to rev the engine and take several minutes to assess the noise. Some WOF stations now have a simple noise meter to do a 'quick check'. Assuming that the inspector is trained to perform this check, it will take more than a few minutes to complete, thus placing the inspector in an even more difficult position. Of course, if the one 'trained' inspector is out to lunch at the time, the test will not be done.

Subjective difficulties.

Assessing sound levels subjectively is notoriously difficult. Doing a noise subjective check is different from the other subjective decisions to be made such as tread depth, windscreen cracks, brake wear, corrosion and so on. The character of the sound will influence the decision, whether louder or not. Having a bad day will come into it. Individuals will make differing decisions.

An individual can just pick a difference in loudness of 2-3dB for sounds of a similar nature, if there is an instant comparison between one and the other. But in the testing situation, there is no comparison available. In this case, a change of 6 to 10 decibels will be needed before an inspector might say that this car is louder than another of the same type, based on a memory of his prior experiences. Given that issue, inspectors will almost certainly be erring on the side of passing, rather than failing a car for noise. Some excessively noisy cars may not be failed. This will tend to maintain status quo and not lead to an overall reduction of vehicle noise in the community.

To improve this situation, noise meters should be mandatory to confirm suspicion of a noisy vehicle. The Australian philosophy of recording noise signatures (levels) of models entering the fleet should be adopted here also. This will aid measurements of the existing fleet over time as signatures for all models are available on the national database.

Not testing heavy vehicles

Since the transport agency considers that heavy vehicles; trucks and buses, are not generally modified, these are not required to take an ONT at any stage. This is more an issue around the importation standards being too lax, however not requiring heavy vehicles to be tested means that it is not recognised that, as they age, they become noisier.

The 'objective' noise test

The ONT is a tailpipe style measurement based on standard ISO5130. The actual test is performed by an LVVTA certifier. Test venues are outdoors. The mic is placed on a tripod 500mm away from the exhaust at a 45 degree angle and the maximum dB(A) value is recorded when the certifier brings the engine up to a specified rpm (different for various vehicle types), then releases the accelerator to bring the engine back to idle. The car is not in gear for this test. There are minor variations to the setting up process for vehicles with twin exhausts and other options.

The full test process (ref 21) can be found at: LVVTA exhaust noise emissions (pdf)

This is fundamentally not a difficult test to set up, although the environmental requirements have to be met. There has to be sufficient open space around the vehicle without proximity of other buildings, objects or people. Clearly, it cannot be conducted when raining, or when there is too much wind because the microphone will pick up wind noise. There is not sufficient sophistication in the standard to permit an indoor measurement. These restrictions can result in some inconvenience to owners forced to present their vehicles for testing, quite apart from the possibility of having to travel some distance to get to the nearest certifier.

Problems with this kind of test

The test is designed to be simple and relatively inexpensive. Naturally this leads to compromises. Does this method accurately reflect the noise of a vehicle as it drives past your house? This is the primary question, along with "Will this standard lead to a statistical reduction of traffic noise?" I see the issues as follows:

  1. It is not a drive-by test.
  2. The vehicle is not measured while accelerating.
  3. The vehicle is not under load during the test.
  4. The single result does not fully represent 'annoyance' neither does it properly characterize perceived loudness.
  5. A vehicle measuring right on the limit is given the benefit of the measurement uncertainty, rather than the other way around which would better lead to lowering of traffic noise levels.
  6. The use of "A-weighting" means that the low frequency energy; boom, roar and rumble are not included in the result.

Taking these points, one at a time:

Stationary test

The stationary tailpipe test largely ignores all sources of noise other than from the exhaust pipe. Although much of a vehicle noise is regarded as being from the exhaust pipe, as much or more noise emanates from the engine bay itself. The LVVTA for example, adds a 4dB allowance when measuring rear-engine cars because of the proximity of the engine to the microphone. This appears to mean that more noise power comes from the engine area than from the tailpipe. The noise limit for motorcycles is adjusted upward because of the same problem. These difficulties alone make the tailpipe test very compromised and unrepresentative, quite apart from the fact that road noise, transmission noise and other noise sources present when a car is in motion, are not measured at all.

Certain allowances are made for this, which is why the noise permitted for the tailpipe test are much higher than limits in a drive-by test. Some correlation has been demonstrated in the past which has led to the limits for the tailpipe test being about 14dB higher than is mandated for type testing using drive-by methods, however that correlation is a moving target with modern vehicles and as new technologies such as hybrids and pure electric vehicles (without tailpipes) make the relationship different. The very fact of the difference 'only' being some 14dB means that most of the noise of a moving vehicle emanates from other than the tailpipe, since, if all the noise from a car was a point source at the tailpipe, the tailpipe test would return a figure over 23dB higher than that measured at the distance used in the drive-by method.

Ratio:(static noise/drive-by noise)=20*log(7.5/0.5) = 23.5dB. (assuming point source)

What partly explains this anomaly, is the fact that type testing, using drive-by methods does include runs made by accelerating past the test microphone as well as constant speed runs. The maximum value of noise under the various conditions is used as the 'noise' figure of the vehicle. Because acceleration under load, especially at higher revs, produces more noise, the drive-by test is more representative of noise levels in practice.

Of course we know by observation that for some vehicles, exhaust noise is entirely dominant and annoying, while for other cars, the exhaust note is barely heard above the road noise. This relationship can go either way and depends very much on how the car is being driven at the time. All this makes it unlikely that a single figure derived from a stationary test can be representative of the potential level of annoyance due to vehicle noise.

Vehicle is not tested under acceleration

As noted earlier, the tailpipe test is done by leaving the gearbox in neutral, bringing the revs up to a specified number, starting the measurement, then releasing the accelerator to get back to idle. By not measuring during acceleration, resonances, rattles and simply the varying effectiveness of the muffler with variable pulse load periodicity, are not being measured.

Vehicle is not tested under load

Engines (internal combustion) produce more noise under load than they do in neutral. Some tests have demonstrated that the noise with and without load tends to converge at maximum revs. However this is unlikely to be consistent in all cases and the static tailpipe test is not done at maximum revs. So, the worst case noise will not be seen. Ideally, if a single figure limit was desired, the limit would be reduced by about 2dB to account for this alone. Rattles, resonances and rasping sounds that occur when under load will not be recorded.

Single figure noise level not representative of loudness or annoyance

Clearly there is some relationship between noise level and annoyance because a completely silent vehicle would not be annoying. However, a single figure, especially one excluding low frequencies will not be fully representative.

Loudness is only loosely related to the 'noise level' as measured by any of the current vehicle noise testing methods. Better methods of quantifying 'loudness' do exist. The loudness of broadband sounds can be estimated by means of the Stevens Loudness Method (ISO 532A). In this method the sound energy is first divided into octave or 1/3 octave bands. A loudness value for each band is first measured, with the total loudness then determined by a summation formula. Another method for estimating total loudness is the Zwicker method. The Zwicker method is also based on the use of octave or 1/3 octave band analysis of the sound signal. The Zwicker loudness process is more complex than the Stevens method because masking effects are accounted for. The loudness value of a given sound is only one aspect of determining an annoyance index.

It is known that other characteristics also influence how annoying a sound is, for example, the degree and rate of change in loudness, the presence of tonal components in the sound, and the sound character as reflected in how sharp, rough or harsh it sounds. Moore and Glasberg proposed additional improvements including a method of calculating the loudness of fluctuating sounds.
See AES convention report at: Link to
Both Zwicker and Moore & Glasberg methodologies have formed the basis of the ITU, CBS and EBU standards which try to quantify loudness of broadcast programme material, and which are embodied in ITU BS.1770-2 and EBU R128. See also ref.25.

Measurement tolerance and repeatability

The local authorities here and presumably in most similar jurisdictions place quite onerous demands on measurement accuracy and repeatability. Noise meters have to be Class 1 or better, have regular traceable calibration, plus undergo on site calibration after every set of measurements. Readings are taken to one decimal place and three readings are required to be within 2dB before being averaged to give the single number.

It is, of course logical to require accuracy of any measurement, but the accuracy required has to fit the purpose. For example, if you asked me how far it is by road from the office to the Wellington central railway station, I could spend half a day with Google Earth or trace the distance out on an accurate map and come up with an answer of 6,108.75 metres. Or, I could recall the last time I did this run in the car, and tell you it is 'about 6km'. You had no need for the better accuracy of the first answer. The more accuracy you need, or think you need, the more you are going to pay, and the cost is sometimes considerably more.

Appropriate accuracy of a measurement is an issue all engineers encounter. Those in administration, or those who are not able to calculate the accuracy needed and the uncertainty of the measurement, tend to err on the side of demanding too much accuracy, although that can be mitigated once they know the cost. Because of the high cost due to accuracy demands, authorities then have little appetite for acquiring sufficient measurement equipment, leading to the downstream effect of needing to simplify requirements in order to match the available measuring equipment. In the case of these vehicle sound levels, authorities are also concerned about challenges to the results.

For sound level meters a basic instrument accuracy of 1dB is more than enough because it is hard to repeat sound level measurements to better than 2dB accuracy even with the greatest of effort. This is especially so in the outdoor environment. It turns out that for most measurements of any kind, operator technique is responsible for greater inaccuracies than is the instrument itself. Presumably those who would challenge a result are those whose cars have 'just failed'. Vehicles which measure that close to noise limits need to be cured, otherwise the measurement uncertainty will not help lower traffic noise.

The practice of making three measurements which must agree within 2dB is reasonable, but to improve the confidence, more readings could be averaged. For example, using three readings of 93dB, 95dB and 97dB (2dB tolerance limit), the average is 95dB and the 95% confidence range is for the mean to lie between 93.2 and 96.8dB. Instead, by taking 5 readings resulting in 93dB, 94dB, 95dB, 96dB, 97dB, the average is still 95dB, but the 95% confidence range is reduced to between 93.8dB and 96.2dB. This allows less wriggle-room for a challenge to the results.

Another notional system might have two thresholds. Using the current N.Z limit for pre-1985 vehicles for example, exceeding 95dB means the vehicle has to cured immediately so that it does not exceed 93dB (2dB below the limit). A vehicle measuring between 93dB and 95dB would be required to be cured by the time of the next WOF (6 months) and it must be retested to be below 93dB also. This would help with both limitation of potential challenge and reduce the burden of accuracy somewhat. The benefit of uncertainty is now skewed towards failing a vehicle rather than passing it, aiding the reduction of community vehicle noise over time.

The (mal)practice of using A-weighting
(6 good reasons to avoid using it)

A-weighting of sound pressure level measurements is almost universal amongst statutory agencies across the globe. It's use is akin to the colossal momentum of a tsunami and almost impossible to stop. Once embedded in the culture as it is, nobody is ever brave enough to move away from it. One agency will use it because another is using it. They will copy the methods because they think that it must be right; being in use practically everywhere. Acoustic consultants understand the limitations but are often compelled to toe the line.

A-weighting is almost completely unsuitable for vehicle noise testing; at least in isolation.

The weighting devalues the contribution of bass (low frequencies); the A-weighting curve does so progressively below 1000Hz. Frequencies of 95Hz contribute only one-tenth of their actual value and going lower still, a frequency of 20Hz is attenuated by 50dB (to 0.3%). The A-weighting curve and siblings, naturally called B and C, were derived from the work by Fletcher and Munson which resulted in their 1933 publication, of a set of equal-loudness contours for the sensitivity of human hearing.

Reason No.1 to avoid A-weighting

Unfortunately, the A-weighting curve, being derived from the 40 phon equal-loudness contour, is only truly appropriate to aid analysis of quiet sounds. My copy of 'Master Handbook of Acoustics' (F. Alton Everest) states that when sound levels are around 80dB SPL and above, the C-weighting curve is to be used. It turns out that the human ear gains much of its bass sensitivity back when the sound levels are high and accordingly, the C-weighting curve has only 6dB attenuation of 20Hz frequencies (leaving 50% contribution). Although the sound levels at a distance from the noise source become lower, they still are not low enough to justify the use of A-weighting. There are additional reasons to include bass energy in results.

Reason No.2 to avoid A-weighting

Bass acoustic frequencies are not easily stopped. A low frequency has a long wavelength. You can put high fences between you and the road, have trees and shrubbery in the way, double-glaze your house windows, but the bass gets through. In general, bass is all that is left if you have made a good attempt at installing acoustic barriers. That is why the boom and rumble of passing vehicles is still present and also why you only hear the thud-thud from excessively loud car stereos. It is because the higher frequencies of the spectrum are relatively easily stopped; but the bass is not. Most domestic housing (especially here in N.Z) has a light timber frame with timber or panel type cladding. Even with fibreglass insulation in walls, and double-glazing, most of the bass energy passes straight through. In fact, bass energy arriving at the outside wall of a house, will vibrate that wall and the windows, so the bass couples to the interior very nicely. Rather like being on the inside of a kick drum. The lower the frequencies, the more easily bass gets through. Even large scale noise barriers on arterial roads have very limited effectiveness at stopping bass. A thick solid concrete or block wall might in principle give some bass stopping power, but the bass goes over the top. Such a wall would need to be prohibitively high. Acoustics engineers use a combination of reflection, absorption and scattering techniques to make acoustic barriers; however even in clever combinations, the methods have limited bass stopping power. Many research projects have shown how much acoustic loss is provided by various wall materials, and they all demonstrate how hard it is to stop the bass.

A quick experiment demonstrates this problem. A series of chirps (frequency sweeps) was generated inside a lounge. The microphone was placed on the balcony outside, which is via glass French doors. Three locations were captured and averaged for the plot below. This is typical of the usual trends seen for frequency dependent losses. Reasonable loss occurs for the higher frequencies, but very little loss to bass frequencies.

Plot of frequency-dependent
acoustic losses through glass doors
acoustic loss through glass graph

Reason No.3 to avoid A-weighting

Resonance of the low frequency (bass) components in a house. Once the bass is within a room of your house, it can be magnified in volume at certain frequencies. The width, height and length of your room will determine which frequencies are magnified. Different rooms will magnify different frequencies. Where you are in the room matters too. Usually the worst place to sit or sleep will be near corners, but other spots within a room also have resonance anti-nodes; these will shift location at different frequencies. Only frequencies up to about 200Hz cause resonances.

The standard 2.4m stud height in modern houses will cause a resonance at 71Hz (and multiples). A 4-cylinder car driving past at, or accelerating through 2130rpm generates 71Hz. A six-cylinder car would have to be pulling 1420 rpm to generate 71Hz. This is a common nuisance frequency because that is a standard room height.

My own lounge is 2.4m high by 8.4m long and 6.2m wide. The primary resonances for this room are 71Hz, 20Hz, 27Hz (and multiples of each). Other resonances will occur from multiple wall bounces but the above will be the strongest. As an example, we always know when my neighbour arrives home in the Subaru Legacy, even though we don't see him or hear the car, except for a very low powerful vibration in the lounge which is felt rather than heard. It occurs for a few seconds as he parks the car. The 20Hz resonance of the lounge just happens to correspond to 600rpm, which for the Legacy, is the idle rpm. The 27Hz resonance is just as likely to be problematical when nearby 4-cylinder cars idle at 810rpm.

For people who are troubled by hard to explain low-frequency rumble in their dwelling, it may help to find a location in the room where the rumble is reduced. It may be necessary to move away from corners or even away from walls. If that doesn't help, there are products loosely called bass traps which once installed in a room can help soak up excess bass. They have to be designed for the actual problem frequency and will have little effect at other frequencies. The only problem with these is that they will change your décor and to be effective at the lower bass frequencies, have to be quite large.

Reason No.4 to avoid A-weighting

A-weighted noise figures have only weak correlation with annoyance

Laboratory tests statistics of time-varying loudness model predictions, particularly the 'loudness exceeded 5% of the time (N5)', have been shown to be more highly correlated to subjects' annoyance ratings of transport noise signatures than metrics based on the A-weighted sound pressure level. In most annoyance modelling for environmental sounds, focus has been on metrics that quantify the noise levels, which is often interpreted as the measure of the loudness of the sound. It is known that other characteristics also influence how acceptable a sound is, for example, the degree and rate of fluctuation in loudness, the presence of tonal components in the sound, and the sound character as reflected in how sharp, rough or harsh it sounds. Acoustic factors also affect how a sound is perceived. Low frequency sounds; being below about 200Hz, are known to contribute to annoyance.

Several laboratory studies have been conducted to determine how annoyance is affected by low frequencies. Persson and Björkman (1988) conducted a study using broad band fan noise centred at 80, 250, 500, and 1000 Hz. They found that the noise centred at 80 Hz was found to be more annoying than the other three sounds. In terms of road noise Nilsson (2007) assessed the annoyance and perceived loudness of road traffic sounds with low, medium, and high levels of low frequency noise. Subjects were asked to assess the annoyance and loudness of the recordings on a magnitude scale. The perceived loudness and annoyance were compared to the A-weighted sound pressure level of the sounds. This comparison indicated that sounds with a high level of low frequency noise were considered both louder and more annoying than sounds of the same A-weighted sound pressure level with low and medium levels of low frequency noise. Another factor that can impact annoyance is the tonality of the sound. Annoyance does increase when a tonal component is present. See references 2,3,4,22,23,24.

Reason No.5 to avoid A-weighting

Much research demonstrates health impacts from low-frequency noise

Symptoms of annoyance due to LF noise are those associated with stress. These include feelings of irritation and unease, fatigue, headache, nausea and disturbed sleep. Sleep disturbance, especially with regard to time to fall asleep and tiredness in the morning, are commonly reported in case studies on low frequency noise. See references 2,3,4,17.

As people age, their hearing sensitivity reduces, but only for mid and high frequencies. Older people become more acutely aware of low frequency noises. Some studies also show that one cannot become accustomed to low frequency sounds in the same way that occurs with higher frequency continuous noise.

Reason No.6 to avoid A-weighting

Those who design and fit after-market muffler modifications generally sell on the basis of increasing engine power (a myth), and additionally try to sell on some subjective basis such as appearance, or the creation of a more 'pleasing' sound. In order to decrease back-pressure on an engine, a more straight-through muffler design is required and that in turn, means that more of the low frequencies; comprising engine and cylinder firing rates plus harmonics, will be let loose. After-market exhaust suppliers know by how much they can increase vehicle noise by, while remaining just under legal limits. They also know that because of A-weighting, they can increase the boom notes considerably without risk of going over those limits.

Spectrum of sound from 2008 Honda Civic 1.8
The spectrum below was generated from a microphone positioned as per ISO5130, at 500mm from the tailpipe.
The Honda engine is steady at 3000 rpm.
Audio spectrum of Honda Civic at 3000 rpm

At 3000rpm, the cylinder firing rate is 25Hz and the engine firing rate (EFR) is 100Hz. These and associated harmonics form the primary engine firing audio components. There is broadband noise, with some predominant spikes over the range 1000Hz to 6000Hz, with a peak around 2000Hz. All these components define the 'character' of the sound. Vertical scale here is in decibels(dB) although the values are correct only in their relativeness. I have not calibrated the microphone to represent sound pressure levels. An A-weighting reading for this car would ignore all the engine firing components, including the peak at 100Hz. In this case, not too much harm is done because the Civic has a relatively civilised engine note without any real boom.

Here is a short segment of the Civic recording (left) plus the same file after being A-weighted (right):

Honda Civic at 3000rpm

Honda Civic at 3000rpm A-weighted

Summarising the noise testing section

Noise from road vehicles needs to be reduced. There is simply no reason for residents to tolerate high noise levels. Most authorities do recognise that there is a problem, but do not assign enough importance to it. The EU has discovered that their lowered noise limits introduced 20 years ago did not have any real impact on traffic noise, so have embarked on another reduction of the limits. New EU limits are at least 10 decibels lower than those recently introduced in N.Z and some Australian tests have shown that new unmodified cars are capable of achieving those lower noise limits. I think that the discussion above goes some way to explaining why the current testing techniques have not had the desired result.

For ease of deployment and low cost, authorities are using simple noise meters; the principles for which have remained unchanged for over 50 years and they are using them in a discredited fashion (the A-weighting). In the 21st century, more sophisticated noise analysis is readily available for little more money and in a form which could be used by non-experts.

At the end of the article is a road-side recording of passing vehicles plus short discussion of results.


Some vehicles pass by sounding like the space shuttle blasting off, while others will pass by with a whisper. And that is not even considering electric vehicles. So, the answer to the heading question is, of course they can. If all cars were not significantly noisier than the quietest example, we would probably not have an issue at all. Of course, the same goes for motorbikes, trucks and buses.

Some manufacturers are making efforts to quieten their new models. They want to have 'a point of difference' in a competitive market and it will benefit all of us if the fashionable point of difference is quietness. Making car interiors quieter will have a simultaneous benefit in the reduction of emitted noise. See a marketing page for the Mazda Premacy. Below; the screenshot from that page:

Premacy ad Note that in addition to the exhaust noise problem, Mazda is looking at acoustic insulation in the engine bay and engine peripherals such as oil pumps and chains. Low noise tyres of course. Some references state the difference of noise between the quietest and noisiest tyre is around 5 decibels. Tyre noise tends to come into play above 50km/h and naturally the road surface plays a part. But all else being equal, a 5dB reduction of road noise is significant.

I also note from the above graphic, that Mazda are clear to state this is for the 'Japanese' model. Japanese vehicle noise standards are similar to U.S standards, which are more strict than N.Z, but not as strict as the recent EU standards. This is a reminder to us that any vehicle export model can be built to given standards that would apply in the destination country. Any jurisdiction with 'loose' noise standards will not get cars fitted with best practice noise reduction.


Rather a lot of science goes into making an effective muffler, although in the interests of cost, not every muffler makes use of all that science. In general, sound waves propagating along a pipe can be attenuated using either a dissipative or a reactive muffler. A dissipative muffler uses sound absorbing material to take energy out of the acoustic motion in the wave, as it propagates through the muffler. Reactive silencers, which are commonly used in automotive applications, reflect the sound waves back towards the source and prevent sound from being transmitted along the pipe.

A reactive muffler usually comprises a series of expansion chambers plus resonating chambers designed to reflect and scatter sound back towards the engine. This kind of muffler is good for automotive applications, where the speed of the engine varies. Varying engine speed causes the airflow speed to vary, as well as the frequency of the engine noise. Mufflers do not have a constant acoustic loss at all frequencies. Unfortunately, any restriction to airflow results in back-pressure, which in turn, reduces engine power. So, compromises are inevitable. More effective noise reduction tends to create more back-pressure. For more detail on muffler design, a good article can be found at:

In all muffler designs the tailpipe length can have an important effect. The tailpipe itself acts as a resonant cavity that couples with the muffler cavity. The attenuation characteristics of a muffler are modified if the design tailpipe is not used. This is one reason why modified cars sound completely different (and always noisier) than the stock versions.

Some after-market muffler kits are sold on the premise of 'increasing engine power'. Well, there is no way for any muffler to increase the power of an engine. However, a design with less airflow restriction will tend to result in reduced loss of peak power. These more 'straight-through' designs rely more on absorption and scattering of sound rather than reflection, and will almost certainly have more low-frequency (boom and rumble) content than the more typical reflective muffler. However, not all of these mufflers are created equal and tests have been performed demonstrating very minor increases in power at the expense of much more noise. (ref.1) It seems, that even in this field, the marketing hype overtakes the reality and those who would spend big sums on exhaust modifications are being deceived about what can be achieved.


In general, more effective silencing requires a physically larger muffler, or a series of mufflers. Space is often the problem, along with cost. Space for effective mufflers is a problem on small cars and of course on motorbikes. An effective and small muffler is more expensive but, as can be observed every day, some motorbikes can be much quieter than others.

Other factors influencing muffler design are mounts and pipe thickness. Mounts have to isolate the muffler and pipework from the vehicle, so to not couple noise and vibration to the chassis. Pipes have to be thick enough to not let noise escape. I had for a time imagined that a concentric piping arrangement might be noise-effective, but I see nobody using this approach. The thought had arisen from considering jet engine technology where most of the engine intake air bypasses the jet core and in the process, helps muffle the jet exhaust noise as it exits out the back. Of course, getting high velocity air into this notional 'outer' exhaust pipe is a problem for a road vehicle and the other issue is that one does not want to rapidly cool the exhaust gases in the pipe anyway.

Space for effective mufflers is not and should not be a problem on heavy vehicles; trucks and buses. There is more than adequate room for large and multiple mufflers, even for side-branch resonators to take out primary engine frequencies. But how often do we see huge trucks with a short vertical pipe and simple absorptive muffler? Engines today have ever-increasing power per unit engine capacity. A small loss of power in order to achieve good silencing is no longer a barrier. New generations of hybrid buses should be able to achieve low noise levels since the internal combustion engine is usually a smaller diesel, running at near constant rpm. Effective silencing of these is not difficult.

Can a car be TOO quiet?

The city of Wellington, in N.Z is often cited by locals as being the jay-walking capital of the world. By observation, there is a clear problem, but I doubt whether it is worse than other similar urban centres around the world. Since I see manufacturers of hybrids and pure electric vehicles looking at methods to 'add' noise to their vehicles, the discussion has obviously arisen elsewhere. These efforts to add noise should be dismissed as needless. Is life so easy that our modern urbanites have lost all survival instincts?

Is life so easy that our modern urbanites have lost all survival instincts?

We have a further issue in Wellington, of people being hit by buses. Pedestrians are not using crossings at the time; neither do buses mount the footpaths. The frequency of these events is increasing. If a person can fail to see an 8-tonne; 4 metre high bus, bearing down on them, then perhaps there is an concern. Noise isn't the problem however, because these same 8-tonne buses make a considerable; sometimes painful level of noise, especially the diesels. It should be stated that this is not to say people should never make a mistake. To err is human and I am well aware of this. In fact some of my professional life has been about discovering whether products, processes and systems are as mistake-tolerant as they should be. Being hit by a bus, or any vehicle legally travelling on the road is not a mistake. It is a fundamental degree of carelessness, which should itself be addressed in the population.

I have seen it written that the reason we should make hybrids and electrics noisier is for the benefit of the blind or partially-sighted. While at first this seems reasonable, you will find that people so afflicted have an enhanced sense of survival and will not step out onto a road carelessly or be relying on their hearing to detect oncoming vehicles, even if their hearing has become acutely tuned.


A large percentage of residents, particularly in urban areas all over the world, rate noise from road vehicles as a most serious concern. Statutory authorities do acknowledge a vehicle noise problem exists, but do not allocate enough resources to either monitoring noise or fixing it. Vehicle noise needs to come down at a rate much faster than by nibbling at the problem over a 20 year cycle. It will cost money, but the amount will be but a fraction of the costs of healthcare for those suffering health effects due to such noise.

Elevate the importance of having a low noise vehicle fleet

This is critical. It needs to be done at the level of the law-makers through to individual vehicle owners. Law makers need to acknowledge that vehicle noise contributes to health problems and that costs of healthcare would reduce if vehicle noise was reduced. To simply make a few minor incremental changes is not enough as even these take years to have any impact.

To improve the 'civility' in our civilisation, action to reduce vehicle noise is needed now. In N.Z, the first action should be to remove the provision to allow owners to modify their cars to make them louder than the original manufacturer supplied them; up to the present exhaust test limit. Modification is fine. Louder than original; no.

To improve the 'civility' in our civilisation, action to reduce vehicle noise is needed now.
The importance of low noise in transport needs to be bedded in with statutory authorities, councils, transport agencies and testing station technicians.

Vehicle owners need to have impressed upon them that making excess noise is not socially acceptable. It is as unacceptable as driving while under the influence of excess alcohol or drugs, or texting while driving, for example. Those who enjoy making a lot of noise and consider that it is their 'right' to do so, would do well to recall the memorable Charles Dickens quotation "The good of the many outweighs the good of the few, or the one". This quotation was also used in one of the Star Trek movies. Nobody has the right to blast around suburban streets in a loud vehicle, disturbing, annoying and contributing to ill-health in possibly, hundreds of residents.

Councils and large fleet operators should make noise emissions part of tender requirements. It is not sufficient to simply say that noise must not exceed current regulations, because as has been discussed, the current regulations are inadequate and due to testing methodologies, do not address excess levels of low-frequency noise from vehicles at all. Develop in-house noise specifications in conjunction with manufacturers. This need not be an onerous or expensive exercise. Some vehicle manufacturers may have to do some further design work, which will have an initial cost, but those costs will amortise very quickly. Other manufacturers may simply have to leave in the noise reduction features that they would otherwise remove for countries who do not demand them. No manufacturer would risk losing a large tender for want of compliance with a noise clause.

Develop better testing methods

At the very least, the transport agency needs to gather noise signatures of all new models entering the country as well as recording signatures for existing models as they pass through testing stations. Taking the example of Australia will be a start. Ideally those signatures ought to be octave, or better still, one-third octave analysed and saved as a digital recording for future analysis, as opposed to a simple A-weighted noise reading.

Below is an octave analysis of the Honda Civic recorded at 3000 rpm. This is the same spectrum shown earlier in this document but with the total energy split into 9 octave bands. The last (pink) band is the rms sum of all the bands.

Octave bands for Civic recording Note the highest level is the 2000Hz band, which we
would have inferred from the full spectrum plot earlier.
The engine firing rate of 100Hz is included in the third band.
A more sophisticated version of analysis like this is a
starting point to produce an annoyance index.

Development of better testing methods as discussed earlier.

Much work has already been done on the subject internationally and it will just be a matter of distilling that work into a methodology. A little post-processing of measurement data can be done in a few seconds and will not impact the time it takes to noise test a vehicle. While it is acknowledged that the provision of full nationwide drive-by testing facilities is not warranted in New Zealand as long as there is no internal vehicle manufacturing, it is an option to provision at least one such facility to help develop and verify vehicle noise testing methods. Abandoned airports or even a small corner on a military air base, for example, are good places to set up such a resource.

Monitoring and enforcement

Regrettably, there is always a need for enforcement. Some states of Australia are starting to listen to the residents and are installing vehicle noise-monitoring facilities at key intersections of certain cities. These would trigger a photograph of the offending vehicle. It is a start. One could imagine that the variables involved would make it difficult to work out which vehicle was causing the noise as well as make it problematical to determine a trigger point. However, over time, the average and typical noise levels will become defined, so that only exceptions will be recorded.

Make it law that after-market exhaust modifiers have to provide a noise certificate for the modified vehicle. Include in the law, penalties for workshops who have misrepresented their work.

Make it simpler for residents and drivers to report a noisy car. At the moment there appears to be no such provision in New Zealand. Any driver can phone in a report of erratic or dangerous driving, but there is no mention of whether that might include noise. The police also have an on-line form for anyone to provide details of bad driving behaviour they have witnessed, but again, it is not clear whether noise complaints will be considered. It appears somewhat onerous to have the police act on a complaint of a loud vehicle. The person making the complaint has to demonstrate that the particular example is noisier than a standard vehicle of the same type. Needless to say, the process is too difficult for most people to act on. One can easily argue that police always have higher priorities, however, the lack of an effective and simple reporting scheme works against any intent to bring down vehicle noise levels.


  1. Review of Methods Used For The Control Of Excessive Noise From New Zealand Road Vehicles. Malcolm Hunt Associates, Noise & Environmental Consultants, Wellington. April 2005
  2. Investigating low frequency noise complaints. Casella Stanger_2001 for DEFRA(UK)
  3. Inadequate standards currently applied by local authorities to determine statutory nuisance from LF and infrasound. Hazel Guest, UK Noise Association.
  4. Addressing Wind Turbine Noise. Daniel J Alberts
  5. Engine Exhaust Noise Control. Jerry G. Lilly, P.E. JGL Acoustics, Inc.
  6. Integrated Assessment of noise reduction measures in the road transport sector. P.A Morgan; P.M Nelson and H. Steven.
  7. More Sound Quality Metrics; Trevor Cox, University of Salford
  8. National Stationary Exhaust Noise Test Procedures for In-Service Motor Vehicles. NTC Australia Sept 2006.
  9. How 'socially disordered' are the places we live in? Crikey/The Urbanist: Alan Davies 22 Feb 2012
  10. SOUND LEVEL OF MOTOR VEHICLES cep Policy Brief 2012-23 of 29 May 2012
  11. Proposal for a Regulation on the sound level of motor vehicles. COMMISSION STAFF WORKING PAPER - EUROPEAN COMMISSION Brussels, 9.12.2011
  12. Environment News service; Traffic Noise Causes Heart Attacks, Early Deaths Across Europe April 2011
  13. Burden of disease from environmental noise. Quantification of healthy life years lost in Europe. World Health Organisation 2011
  14. VENOLIVA - Vehicle Noise Limit Values - Comparison of two noise emission test methods -Final Report F. de Roo MSc. M.G. Dittrich MSc. P.J.G. van Beek MSc. C. Bosschaart MSc. G.B. Derksen MSc. M. de Kievit MSc. 30 March 2011
  15. Environmental Noise Measurement. Bruel and Kjaer
  16. FUNDAMENTALS OF ACOUSTICS Professor Colin H Hansen. Department of Mechanical Engineering, University of Adelaide.
  17. Low frequency noise and health effects. Mariana Alves-Pereira, Lusofona University, Lisbon and Nuno Castelo Branco; Centre for Human performance, Alverca, Portugal. June 2011.
  18. New EU Vehicle noise limits Briefing paper. EU Transport and Environment. April 2002
  19. N.Z Land Transport Rule Vehicle Equipment (Noise) Amendment [2009]
  20. NZTA Guide to owners who have failed a WOF noise check
  21. N.Z Low Volume Vehicle Standard 90-20(03) (Exhaust Noise Emissions) Low Volume Vehicle Technical Association Incorporated
  22. Noise Pollution: A Modern Plague. Lisa Goines, RN and Louis Hagler, MD
  23. Study of Vehicle Noise under Different Operating Conditions: Ashish BASKAR, EPFL - LAVOC, Dr. Edward CHUNG, EPFL - LAVOC, Prof. André-Gilles DUMONT, EPFL - LAVOC, Conference paper STRC 2006
  24. An overview of methods to quantify annoyance due to noise with application to tire-road noise Sarah Mcguire, Patricia Davies
  25. Sound Quality and Psycho-acoustic metrics: LMS Engineering Innovation

Some material from the above referenced items is quoted and also short excerpts have been directly used in the cases that no better phrasing was possible.

APPENDIX: Road-side recording
and brief discussion

I undertook an 6-hour basic roadside recording of traffic noise outside the office. The idea was to record passing traffic from near the roadside and simultaneously record the sound inside a room of the office, close to the road. This would highlight the effects of low frequency noise inside a room. Thanks to Audacity's new VOX feature, I did not have to babysit the equipment all day nor have to cope with a 30GB file at the end. The recording is split into 25 events all separated by a half-second silence. Below is a shot of the audio waveform:

waveform cap

The recording does include overflying aircraft; events 2 and 8, plus a helicopter; event 24, and two instances of car doors being closed; events 12 and 15. I left these five in for interest but they are not analysed in the table of recorded sound levels below.

To listen to the recording please click the audio slider. The recording is 3min 55secs and is compressed to be about 2MB in size.

Roadside recording: L=outdoor mic; R=indoor mic

About the recording

The outdoor mic is set about 7 metres from the road, clear of other large structures. The indoor mic is set in the corner of the closed room which is 4.5m x 3.3m x 2.4m. The indoor mic is some 4.6m further from the road than mic 1. Sensitivities of each mic have been corrected in post processing, so the relative levels remain correct. Initial sampling rate was 44.1kHz/16 bit and was saved in the uncompressed 'wav' format for analysis. The file was later downsampled to 22.05kHz and compressed to make the published audio track. The intent is to be illustrative and so there are some limitations. The outdoor recording has minor restrictions of dynamic range. The indoor recording has a noise floor limit. Neither mic is calibrated for SPL and so these values are not available, however all dB figures maintain correct relativity.

Roadside recording data

Following is the data table made from the recording.

street recording data tableAs mentioned, the dB figures in this table are not dB SPL, but dB(relative). 0dB has no other meaning than representing full scale (sine-peak) of a 16 bit recording. The rms values in a 400ms window are reported.

Explanation of columnar data

There are some interesting observations to make from this recording.

Some examples of the spectral content are below: Blue trace is the OUTDOOR mic White is INDOOR mic:

E19 spectrum
Significant bass spikes falling on a major room resonance at 40Hz making the indoor level exceed the outdoor by over 9 decibels.

E1 spectrum
Large bass spike at 64Hz.

E20 spectrum
All bass energy with little mid/high notes. Passes type testing because of this despite considerable boom.

E3 spectrum
Large comb of discrete notes plus hash all the way to 8kHz.

E8 spectrum
Included this event to demonstrate how room gain at low frequencies increases the boom indoors considerably.

Axino-Tech Consulting & Services
September, 2012


From: Phillip Leslie-Collins
A big thank you for your article on unwanted noise from road vehicles. Many of the concerns I have about noise pollution in Wellington feature in the article, along with a lot of interesting and pertinent information. Many's the time I've wished I could measure the db level of the 'brake dump' (as the man from Designline called it when I phoned to enquire about the noises their buses made) that scares the bejeezus out of me as a bus pulls up at a bus stop. I totally agree that stress from unwanted or unexpected noise is being largely ignored as a health problem. There's a classic study from the 1970s that I often think about whenever a Newlands bus driver sounds their horn on Willis St or Lambton Quay (well, I think about it once my hair has settled back onto my scalp, my heart rate has slowed and my blood pressure has returned to something like normal). In the study, psychologist J.M. Weiss found that rats who received electric shocks preceded by a warning tone exhibited signs of stress only slightly higher than those in the control group, who received no shocks. By contrast, rats in a third group who received shocks with no warning developed stomach ulcers. Anyway, before I digress any further, thanks for providing such a great resource and for all your research and work on the article. Phillip.
April 2015

Make a comment using the form on the contact page