An overall A-weighted sound level describes the combined sound at the measurement position, but it does not show how that sound is distributed across frequency. Octave-band analysis separates the measured sound into frequency bands so that the assessor can see whether low, middle or high frequencies dominate the spectrum. This information is particularly useful when the purpose extends beyond establishing an overall level. Engineering controls act differently across frequency, barriers and enclosures may perform unevenly, and hearing protectors do not attenuate every band by the same amount. A spectrum therefore supports decisions that cannot be made reliably from a single overall value. Octave-band analysis should be linked to a defined operating condition, position and source or task. A technically correct spectrum collected during an unrepresentative condition can direct attention towards the wrong control problem.
Frequency describes how rapidly sound pressure fluctuates. In workplace terms, lower-frequency sound is often experienced as a hum, rumble or slow mechanical character, while higher-frequency sound may have a sharper or more tonal character. These descriptions are only general indicators; the measured spectrum is needed to identify the bands that actually contribute.
Two sources can produce a similar overall A-weighted level while having very different spectra. One may be dominated by lower-frequency mechanical energy, while another may be dominated by higher-frequency airflow or impact components. Treating them as acoustically identical can lead to control measures that perform poorly.
The spectrum can also reveal a prominent band that suggests a tonal component associated with rotational speed, resonance, airflow or contact between machine parts. Octave bands do not provide the narrow detail of a specialised tonal analysis, but they can show where further investigation should be concentrated.
Frequency data should not be considered in isolation from observation. The assessor should record which equipment was operating, the production state, the measurement position and any temporary events. A peak in a band is more useful when it can be related to a credible source or transmission path.
The instrument and analyser function should be suitable for the intended occupational measurement and should have current calibration control. The microphone position should represent the sound field being investigated, and the assessor should avoid creating reflections or shielding that alter the spectrum.
A stable source may be represented by a suitably observed measurement period. A cyclic or intermittent source may require the assessor to capture and identify the relevant operating phase rather than combining unrelated states without explanation. Where several sources operate together, additional source-specific measurements may help distinguish their spectral contributions.
Background sound should be considered where it materially influences a band. A low-frequency band, for example, may be affected by building services or distant plant even when the target machine is not the dominant source in that band. The report should distinguish measured combined conditions from source-isolated conditions where isolation was possible.
ADOSH-SF CoP 3.0 provides octave-band A-weighting corrections for converting octave-band sound pressure levels to an overall A-weighted value. These correction figures come from CoP 3.0 and apply within the Emirate of Abu Dhabi.
The correction is applied to the measured level in each corresponding band. Negative corrections reduce the contribution of the lower-frequency bands to the A-weighted result, while the positive corrections increase the corrected value in the stated bands.
The table should be applied exactly to the matching centre frequency. Values should not be shifted between bands or applied to a different spectral format. The raw band levels, corrections and corrected band levels should remain visible in the calculation record so that another competent person can follow the process.
Corrected octave-band levels cannot be added arithmetically because decibels are logarithmic. Under the method stated in CoP 3.0, each corrected band level is converted to an intensity unit, the intensity units are summed, and the total is converted back to a sound level.
CoP 3.0 states that where the total falls between two table values, the next higher sound level is taken. This convention should be applied consistently and should be stated in the report rather than hidden within a spreadsheet cell or software setting.
The calculated overall value can be checked against a directly measured A-weighted level collected under the same conditions. A close comparison can support confidence in the data, while a marked difference may indicate mismatched time periods, omitted bands, setting errors, changing source conditions or transcription problems.
Engineering noise controls should address the dominant source, path or receiving condition. The spectrum helps determine which control principle is likely to be effective. A broad high-frequency contribution may respond differently from a low-frequency tonal or structural component, and a material selected for one frequency range may have limited effect in another.
For airborne sound, frequency information can inform the design of enclosures, barriers, silencers and absorptive treatment. The construction, sealing and geometry need to suit the dominant bands. Small gaps may undermine an enclosure, while absorption placed in a space dominated by another transmission path may produce little change at the occupied position.
For mechanically transmitted sound, a spectrum may point towards resonance, vibration transfer or a rotating component that warrants engineering investigation. The acoustic result does not by itself diagnose the mechanical defect, but it helps focus examination and provides a baseline for checking whether the chosen intervention changed the relevant bands.
A control comparison should retain both the overall level and the spectra before and after the change. A reduction in one band may be offset by another source or by a change in operating state, so the complete pattern should be reviewed rather than selecting only the most favourable value.
Detailed control design and protector-selection procedures are addressed separately on /noise-control-engineering and /hearing-protection-selection.
Hearing protector attenuation varies with frequency. An overall rating provides a convenient summary, but octave-band data can support a more specific comparison between the workplace spectrum and the manufacturer's attenuation data. This is particularly relevant where one part of the spectrum dominates or where communication and audibility need to be considered alongside protection.
The assessor should use the manufacturer's data for the exact protector model and fitting arrangement. Substituting data from a similar-looking product can produce an unsupported result. The calculation should show how attenuation has been applied to each band and should retain the unprotected spectrum for comparison.
ADOSH-SF CoP 3.0 states that single hearing protector attenuation is the manufacturer's NRR minus 7 dB. This attenuation figure comes from CoP 3.0 and applies within the Emirate of Abu Dhabi. The Code method should not be mixed silently with a different octave-band method. Where more than one method is presented, the purpose and basis of each should be stated so that the reader can understand why the results differ.
Selection should not be reduced to obtaining the greatest possible attenuation. A suitable protector should reduce exposure appropriately while remaining compatible with the work, other protective equipment, communication requirements and reliable fitting. The spectrum contributes evidence, but workplace usability remains part of the decision.
A frequent error is to treat the loudest unweighted band as the sole control priority without considering the A-weighted contribution, operating duration or source relationship. Another is to compare spectra collected at different positions or under different machine states as though only the control measure had changed.
It is also incorrect to add decibel band values directly. The logarithmic conversion step is essential. A calculation that produces an overall value without showing the correction and combination method may be difficult to verify and can conceal a spreadsheet error.
Frequency labels and units should be checked carefully. Octave-band centre frequencies should not be confused with other band formats, and a spectrum should not be presented without identifying the weighting and measurement condition. Clear presentation prevents later use of the data for a purpose it cannot support.
-26 dB
-16 dB
-9 dB
-3 dB
0 dB
+1 dB
+1 dB
-1 dB
ADOSH-SF CoP 3.0 Occupational Noise, Version 4.0, dated 15 July 2024, sits in the mandatory Codes of Practice layer and applies to all employers within the Emirate of Abu Dhabi. CoP 3.0 includes the octave-band A-weighting correction values and the method for combining corrected bands into an overall A-weighted level. Frequency analysis should be traceable to the operating condition, instrument record and calculation used, particularly where it supports control or hearing protector decisions.
Not necessarily. It is most useful where frequency content affects the question being answered, such as control design, source comparison or protector assessment. A routine overall measurement may not require a spectrum, provided that the data collected are suitable for the assessment purpose.
A-weighting applies different corrections across frequency. A high lower-frequency band may receive a substantial negative correction, while a smaller band elsewhere may make a stronger contribution after correction.
They can indicate the frequency region associated with the sound and may support a mechanical investigation, but they do not prove the exact defect. Confirmation may require equipment inspection, vibration analysis or controlled operating tests.
It is useful where the enclosure design depends on frequency performance. The spectrum helps the designer match mass, absorption, sealing and other features to the dominant sound, rather than choosing materials solely from a general product description.
It can be converted to an overall A-weighted value using the method in CoP 3.0, but the raw spectrum and calculation should be retained. A directly measured overall value under the same condition may also provide a useful cross-check.