Noise Measurement Terms Part 1 – Frequency Weighting & Level and IEC 61672 : 2003

There are many tens of descriptors used for noise but for airport and environmental noise monitoring. Many, indeed most, are not relevant. That is why we are introducing a Noise Measurement Terms service to give descriptions of those parameters commonly used in an environmental monitoring system.

In some descriptions, mathematical accuracy has been sacrificed to aid understanding but there are many ISO and IEC documents that give more precise descriptions. These can be hard to understand unless you have a strong mathematical background. In some cases, where a parameter is especially important, after a simple description we have added the full and formal data, in some cases by giving the formula.

One of the best documents ever published for the definitions of the various parameters is the IEC standard IEC 61672 : 2003 [1]. This was prepared by Working Group 4 of IEC Technical Committee 1, presided over by Alan H. Marsh (USA) and consisting of experts from the top sound level meter companies and some government laboratories. Naturally Cirrus Research plc provided engineers and two staff design engineers, Robert W. Krug (USA) and Alan D. Wallis (UK & New Zealand) had significant input over many years. Indeed, their work forms the basis of this article.

Many workers who do not have a mathematical or science training often find the plethora of metrics and their relationships with each other quite daunting. In this article, we attempt to de-mystify them – just a little!

The headings below fall into several categories and start with things that are not noise descriptors in themselves but are an essential part of them. One problem is that the formal and correct descriptor, description or symbol is often not used, especially by older workers who have used a parameter long before it became internationally standardised. Perhaps the most common is the name for an linearly integrating meter. The correct formal name is an “integrating-averaging sound level meter”, this is the required title to comply with IEC 61672 : 2003, but most people simply call it an “Leq meter”.

Frequency Weighting

The frequency weighting is the first parameter to appear in noise descriptor symbols after the ‘L’ for level. For example see the A-frequency-weighted equivalent continuous level LAeq(t), properly called LAt. In acoustics, ‘Levels’ are always in decibels.

Traditionally A-frequency-weighting was used for quiet sounds, ‘B’ for intermediate ones and ‘C’ for loud noise, but today this is regarded as logical nonsense. ‘B’ is no longer in any International standard and very little practical use, except for entertainment noise. D-frequency-weighting also once existed but this was for use only with non-bypass jets and today these are all military, now Concorde is grounded. It follows that D-frequency-weighting should now no longer be used for measurement of civilian aircraft, international agreements all specify ‘A’. Despite this, some “experts” still ask for D-frequency-weighting.

A-frequency-weighting is specified all over the world as the frequency weighting to use for environmental monitoring for both the exponential sound level and also for the integrated metrics of Leq and SEL (Sound Exposure Level or LAE). A-frequency-weighting has about 0.1% of the sensitivity at 31Hz as it does at 1kHz. While technically, anything that is A-frequency-weighted should be expressed in dB(A), almost everyone simply uses dBA. If no weighting letter is present it is usually assumed that ‘A’ is intended. Theoreticians and purists quite rightly say that A-frequency-weighting is far from ideal for all noise, but not only is there a huge body of existing data, to allow new data to be compared with old; but almost all authorities mandate its use for Health and Safety as well as environmental work.

C-frequency-weighting is specified in the EU for measurements of the Peak value for Health & Safety, but is rarely used in environmental measurements. However, if there is a very high impact noise – a pile-driver and a gun club are examples, C-frequency-weighting may be used in conjunction with a Peak detector or with LAE. C-frequency-weighting has a flat response from 31,5Hz to 8kHz; technically, it is 3dB down at these frequencies. However, today, perhaps a more sensible approach is to use SEL or LAE for such noise sources.

Z-frequency-weighting (zero) is a ‘wider’ version of ‘C’ and is new in IEC 61672. As yet, it is not widely used for environmental monitoring but it will become more and more common for peak measurements. ‘Z’ was ‘invented’ by working group 4 of IEC TC1, as the limit values of ‘C’ (31,5Hz and 8000Hz) are too close together to capture all the spectrum of peak sound level. To acquire all the data when using Peak, manufacturers used to fit “All-Pass”, “Flat” or “Linear”. However every manufacturer used different frequency boundaries as well as different descriptions and their instruments often gave differing readings; Z-frequency-weighting is supposed to take care of this.

Formally, frequency weighting is defined as :- the difference between the level of the signal indicated on the display and the corresponding level of a constant-amplitude steady-state sinusoidal input signal, as a function of frequency and is expressed in decibels (dB).


Most – but far from all – descriptors are Levels and are thus in decibels and this causes the most confusion.

Mathematically, the decibel is NOT a unit as it has no dimensions; it is a logarithmic ratio using base 10 logs (lg). Thus, ANYTHING can be described in decibels and because of this, the parameters used MUST be carefully noted as for example, 10dB(beer) is not the same as 10dB(wine)…

Traditionally in noise measurements the decibel is described as :- twenty times the logarithm to the base ten of the ratio of the root-mean-square of a given sound pressure to the reference sound pressure and it is always expressed in decibels (dB); having the symbol Lp. However, when you are describing ‘energy’ metrics such as SEL or Leq, – as opposed to the ‘pressure’ metric of sound level – the definition is different.

It follows that you CANNOT add two decibel levels, you must first convert the decibel value back into real pressure units – normally Pascal (Pa), by taking the anti-log, and then add the pressures together and take the base ten log of the result. Very easy for a computer or scientific calculator, but by hand it is hard, requiring log tables or a slide rule – remember them? The Cirrus Noise Pole makes this computation many thousand times per second to get such things as the overall Leq.