An acoustic file can be flawless and still disappoint. Maximum ground level several decibels below that of the rotorcraft it replaces, lower cumulative energy per event, comfortable margins on all three measurement conditions: everything is accurate, everything is verifiable. And yet the aircraft is heard sooner, for longer, and draws more complaints. The gap is not a measurement error. It comes from the fact that the metric was built for a single, slowly turning rotor, and a set of small, fast rotors simply does not present it with the same acoustic object.
What the legacy metric was built to catch
Rotorcraft acoustic certification rests on energy-based quantities: effective perceived noise level for the heaviest machines, sound exposure level for the lightest, measured on takeoff, overflight and approach along a reference flight path. All of them integrate the square of the pressure over the duration of the event, then reduce the result to a single number. The choice is not arbitrary: it describes a precise physical object.
A main rotor ten to twelve metres in diameter turns at a few hundred revolutions per minute. With four blades, the blade-passage frequency is a few tens of hertz, and most of the radiated energy sits in that fundamental and its first harmonics, so below two hundred hertz. The mechanisms that feed it have long been identified: thickness noise, linked to the volume displaced by the blade; loading noise, linked to unsteady lift; blade-vortex interaction on descent, which produces the characteristic slap; high-speed impulsive noise when the advancing blade-tip Mach number approaches unity. Each has a short time signature and a marked directivity, often forward and downward. Microphone placement, the reference approach angle and the integration window were all set to catch exactly that.
The whole structure rests on three assumptions: a dominant source, an established regime, a signature that repeats from one flyover to the next. Step outside them, and the quantity keeps calculating without a hitch. It simply no longer describes the same thing.
The spectral shift: small blade, fast rotation
A multirotor aircraft replaces one large, slow rotor with six, eight or twelve rotors one to two metres across. For equivalent thrust, swept area collapses, disc loading rises, and rotation speed climbs, typically by a factor of five to ten. The blade-passage frequency is no longer a few tens of hertz but a few hundred, and the useful family of harmonics extends up to several kilohertz. Added to this is broadband boundary-layer and trailing-edge noise, itself shifted toward the upper part of the spectrum because it follows chord length and relative speed.
This shift is enough to turn A-weighting on its head — a filter that is nothing more than a model of the ear’s sensitivity at low level. It removes close to forty decibels around thirty hertz, about twenty around a hundred hertz, three around five hundred, then nothing at all between eight hundred hertz and six kilohertz — where it even adds a decibel. In other words, the filter used to erase most of a single rotor’s signature; it removes almost nothing from that of a set of fast rotors.
The problem is that helicopter noise metrics were designed for big, slow rotors. Multirotors have many small, fast rotors that make higher-pitched sounds. The standard measurement filter removes low frequencies but lets through the high frequencies from multirotors, making them seem louder than they physically are. It’s like using a ruler to measure something that’s actually a different shape.
A traditional helicopter like the Airbus H125 (12m rotor diameter, 400 RPM) has its fundamental tone at 26.6Hz. A multirotor like the Volocopter VX4 (8 rotors, 2m diameter, 1200 RPM) has its fundamental at 160Hz. When measured with A-weighting, the helicopter’s noise is heavily reduced, while the multirotor’s high frequencies remain almost unchanged. This means a multirotor could have lower actual acoustic power but measure similarly loud to a helicopter in the certification process.
The practical consequence reads both ways. A multirotor aircraft can radiate a total acoustic power well below that of a helicopter and still display an A-weighted level of the same order. The comparison in dB(A) remains honest with respect to hearing sensitivity, but it suggests an equivalence of annoyance that does not exist: it is the mechanisms, not the level alone, that have changed in nature.
Modulation: what an energy average erases
Several rotors never turn at exactly the same speed. A gap of a few revolutions per minute between two of them produces a beat at the difference frequency, hence a slow amplitude modulation of the signal received on the ground. Add aerodynamic interaction — the wake of an upstream rotor ingested by a downstream rotor creates unsteady loading, broadband widening and crossing tones — and the result is a sound whose envelope breathes continuously.

And the ear is an excellent envelope detector. The sensation of fluctuation peaks for modulations on the order of a few hertz, typically around four; roughness takes over for modulations of several tens of hertz. A rotor-to-rotor beat almost always falls in the first band, the one where sensitivity is highest. That is what gives a multirotor flyover its “nagging” character, where a helicopter’s pass is louder but more cleanly finished.
No energy-based quantity has any grip on this phenomenon. The equivalent level, the sound exposure level and the effective perceived noise level all integrate the square of the pressure: they return the same value for a flat signal and for a signal that modulates by ten decibels peak-to-peak around the same average. The only time-related term they carry is a duration correction, based on the time spent within ten decibels of the maximum. It says how long the event lasts, never how the level behaves within it.
The trap. Two aircraft showing the same sound exposure level at the measurement points can produce unrelated complaint rates, and modulation is not the only cause. Standardised measurement conditions cover established regimes, along a reference flight path; the transition between powered lift and wing-borne flight, often the noisiest phase, is poorly represented there — yet it is exactly what happens above the built environment at the entry to a platform, at low height and high load. Comparing two machines on their certified values alone amounts to comparing them outside the point that decides acceptability, and the gap is discovered after entry into service, once flight path, infrastructure and slots are already fixed.
Why the tone correction undercounts multiple tones
The effective perceived noise level does not ignore tonal character: it carries a correction computed on the one-third-octave spectrum, step by step in time. Bands that emerge from their neighbours are sought, a penalty is derived, and only the largest is kept at each instant. The mechanism was designed for the fan tone of a turbojet: a clear, isolated tone sitting on a broadband floor.
A set of rotors presents an altogether different structure, and the mismatch is paid for three times over. First, the width of a one-third-octave band grows with frequency: beyond a few hundred hertz, it ends up exceeding the spacing between successive harmonics. Several tones then fall into the same band, the contrast with neighbouring bands flattens out, and the penalty vanishes — even though the tonality itself remains perfectly audible. Second, the maximum-value rule only penalises one band at a time: ten prominent tones are worth the same correction as a single one. Finally, perceived tonality depends as much on the regularity of the harmonic series as on the emergence of an isolated band; the one-third-octave band does not have the resolution needed to describe that regularity.
| Characteristic | Single main rotor | Set of small, fast rotors |
|---|---|---|
| Blade-passage frequency | A few tens of hertz | A few hundred hertz |
| Band where the energy sits | Below 200 Hz, low harmonics | 200 Hz to several kilohertz |
| Effect of A-weighting | Strong attenuation, several tens of decibels | No attenuation, even a slight gain |
| Time character | Impulsive, periodic, short | Modulated, slow envelope, spread out |
| Tonal structure | Few strong, well-separated tones | Tight tone families, sometimes offset |
| What the tone correction sees | Tones emerge clearly | Tones are drowned within the one-third-octave band |
| Directivity | Marked, forward and downward on descent | More distributed, variable in transition |
| Decay with distance | Negligible air absorption | Significant air absorption |
The ground measurement point is no longer neutral
The certification microphone is set at head height, on the order of 1.2 m, above a reflective surface. Sound therefore arrives by two paths, direct and reflected, whose path difference is roughly twice the microphone height when the source is high. This produces an interference comb: a first dip around a few tens of hertz, then an alternation of dips and peaks every one to two hundred hertz.
For a single rotor whose fundamental sits around twenty hertz, all the useful content lies below the first dip: the measurement is robust, and an error of a few centimetres in microphone height changes nothing. For a dominant harmonic at three hundred or six hundred hertz, the same geometry can place the tone on a dip or on a peak. A few decibels of difference can then hinge on tripod height, a lateral shift in flight path, or ground type. The quantity remains calculable, but its repeatability drops exactly where the content has become tonal.
Atmospheric absorption works in the same direction. On the order of a decibel per hundred metres beyond two kilohertz, nearly zero below two hundred hertz, it depends strongly on humidity and temperature. The high-frequency signature therefore fades faster with distance, which benefits distant residents; but corrections to reference conditions, marginal for a rotorcraft, become a first-order term in the uncertainty budget.
The right reflex. Require that every ground campaign keep calibrated time-domain recordings, not just the integrated values: a loudness, a tonality or a modulation depth can be recalculated later from them, where a table of sound exposure levels no longer allows anything. Log microphone height, ground type, temperature and humidity — without those four lines, no comparison between two campaigns holds up.
Detectability, duration and number of events
Residential annoyance does not start at the maximum level: it starts at detection. The ear detects within critical bands, and a tonal line emerges as soon as it exceeds the ambient floor within its own band — well before a broadband noise of the same total energy becomes perceptible. In a quiet rural nighttime setting, where the floor is low and itself concentrated at low frequencies, a tone of a few hundred hertz remains audible long after the overall level has dropped back to the floor. This is the mechanism that explains why a “quieter” aircraft is heard sooner and tracked for longer.
The second factor is the number of events. An air-taxi service does not target one helicopter every two hours but one movement every few minutes at peak times. Long-term indicators sum energy over the day, with conventional penalties in the evening and at night: dividing an event’s energy by ten and multiplying its number by ten leaves the result unchanged. Awakenings, conversation interruptions and task disruptions, however, grow with the number of occurrences. Hence the value of count-based indicators, which count events exceeding a threshold instead of averaging their energy.
| Indicator | Measures well | Does not see |
|---|---|---|
| Maximum A-weighted level | The peak, the sudden emergence | Duration, modulation, tonality |
| Sound exposure level | The energy of an isolated event | The time shape within the event |
| Effective perceived noise level | Duration and a clear, isolated tonality | Tight tone families, modulation |
| Long-term annoyance indicator | A site’s overall load | The misleading equivalence between number and intensity |
| Threshold-based event count | Repetition of occurrences | The character of the sound at each occurrence |
| The real perceived volume, beyond a single filter | Fluctuation, if there is no time-varying version | |
| Tone-to-noise ratio, prominence | The emergence of a tone within its neighbourhood | The cumulative effect of a harmonic series |
| Modulation depth | The envelope and its rhythm | Nothing about absolute level: to be crossed with another indicator |
The levers that actually move the needle
On the design side, blade-tip speed remains the first-order lever. An unsteady loading source radiates like a dipole, whose acoustic power varies roughly as the sixth power of speed: shaving a tenth off tip speed removes on the order of two to three decibels, a fifth roughly double that, and this with no downstream treatment at all. The price is paid in blade area at constant thrust, hence in mass, inertia and drag — this is where the design trade-off is settled, and it is settled early, because no later adjustment makes up for too small a diameter.
Next come the spectral-shape levers. Increasing blade count raises the passage frequency and reduces the amplitude of each tone, but pushes further into the ear’s zone of maximum sensitivity: the gain is not automatic. Modulated spacing, meaning unevenly distributed blades, splits the tonal energy across more harmonics of smaller amplitude and brings the signature closer to broadband content, which is better tolerated. Rotor spacing and vertical offset between planes govern wake ingestion. Finally, control of relative phase is the only way to act directly on the beat: locking phases to suppress modulation, or steering them to point interference lobes away from the ground.
On the operational side, margins are gained differently. A shallower descent reduces blade-vortex interaction but lengthens exposure and spreads the footprint; a steeper descent does the opposite, and the balance point depends on the fabric overflown, not on any general chart. Spreading corridors across several axes divides the number of events per resident without changing the level of a single pass by one decibel: it is often the most effective lever in the short term. Finally, the transition between powered lift and wing-borne flight deserves separate scrutiny, since it is the noisiest regime and the one least well covered by standardised measurement conditions.
What to check before opening a site
Three checks are worth the time they cost. The fine-band spectrum, not just one-third-octave, over the whole duration of the flyover: it is the only document that shows whether the signature is a dense harmonic series or genuinely broadband content. A time trace of the level, sampled finely enough to reveal the envelope: if the modulation depth reaches a few decibels peak-to-peak at a rate of a few hertz, no integrated indicator in the file will tell you so on its own. Finally, the number of movements per hourly band assumed in the study, to be checked against the intended commercial model rather than the certification scenario.
The grey areas deserve naming. Environmental specifications applicable to vertical take-off and landing aircraft draw on the rotorcraft framework and retain energy-based quantities: they produce a value comparable from one manufacturer to another, which is their purpose, but they do not claim to predict local acceptability. No amplitude-modulation penalty method has become established in certification to date, and sound-quality indicators belong more to contractual or local use than to regulation. The gap between what is certified and what is accepted is therefore settled in the impact study and in the agreement with the local authority.
For anyone preparing a type rating on this kind of aircraft, the consequence is concrete: noise-abatement procedures will not be reducible to a single approach profile learned once and for all. They will combine rotor-regime management, a vertical profile and an axis, and will vary by site — so it is worth getting used, from training onward, to reading an acoustic footprint map the way one reads an approach chart. The subject is already being discussed on the FL330 forum and is gradually entering training curricula.
This article sets out physical mechanisms and measurement principles; it does not replace any binding document. Applicable certification specifications, national and local environmental regulations, and the operator’s approved documentation prevail at all times over what is written here.
Questions fréquentes
Does a lower A-weighted level guarantee fewer resident complaints?
No. A-weighting models the ear’s sensitivity to a steady noise, not the annoyance caused by a tonal, modulated sound. A multirotor aircraft can show a lower A-weighted level than a helicopter while being detected sooner, because its signature emerges from the ambient background within a narrow critical band. The level remains a necessary indicator, but it must be cross-checked with tonality, modulation depth and audibility duration.
Why isn't the tone correction in the effective perceived noise level enough?
That correction was designed for a clear, isolated tone on a broadband floor, typically a fan tone. It is applied to the one-third-octave spectrum and keeps only the largest penalty found at each instant. But a set of rotors produces a tight family of tones which, beyond a few hundred hertz, fall several to a band and stop emerging from their neighbours. The tonality remains audible while the penalty cancels out.
Should rotors be synchronised, or deliberately desynchronised?
Both strategies exist and do not aim at the same effect. Locking the phases suppresses the beat and therefore the slow modulation, but can reinforce certain directions through coherent addition of the tones. Deliberately leaving speeds different spreads the tonal energy but creates precisely the fluctuating envelope the ear follows best. The choice is made in light of the directivity pattern toward the ground, not a general principle.
What supporting data should be requested in a platform's acoustic file?
Three items are decisive. The fine-band spectrum over the whole duration of the flyover, which reveals the harmonic structure invisible in one-third-octave form. The time trace of the level, sampled finely enough to reveal the envelope and measure the modulation depth. And the number of movements per hourly band actually targeted, since long-term indicators treat ten discreet events and one loud event as equivalent.
Does distance to the nearest resident protect a multirotor aircraft more?
Partially, and for a precise physical reason: air absorption reaches on the order of a decibel per hundred metres above two kilohertz while it is negligible below two hundred hertz. A multirotor’s high-frequency content therefore fades faster with distance than a rotorcraft’s low frequency. But the attenuation depends on humidity and temperature, and the low harmonics persist: distance changes the timbre more than it removes detectability.

Leave a Reply
You must be logged in to post a comment.