Why Leaning Into a Microphone Adds Bass and No Treble

This page contains affiliate links. If you buy through one we may earn a commission, at no cost to you.

Your voice does not get deeper, and that is worth starting with, because it points at where the answer is. Nothing about you changes when you lean in. What changes is the microphone's response, by an amount that depends steeply on frequency and on distance, and the amount is large: moving from 12 inches to 2 inches adds about 14.4 decibels at 100 hertz and about 0.3 of a decibel at 4 kilohertz. The low end comes up by a factor of 27 in power and the consonants do not move. That is the boom, and it is not the microphone being bad. It is the directionality you paid for, arriving as a side effect.

Two kinds of microphone, and only one of them has this problem

A microphone has to pick one of two things to measure, and the choice decides everything that follows.

A pressure microphone has a sealed back. The diaphragm feels the sound pressure on its front face and nothing else, so its output follows the pressure at that one point in space. Pressure is a scalar with no direction in it, so a pressure microphone cannot tell where a sound came from. That is an omnidirectional capsule, and it is the simplest microphone there is.

A pressure gradient microphone is open at the back. The diaphragm feels the pressure on both faces and responds to the difference between them, which is to say to the pressure gradient along its axis. A gradient is a vector, so this microphone can tell direction, and all the patterns anyone sells, cardioid, supercardioid, hypercardioid, figure of eight, are made by mixing a pressure response and a gradient response in different proportions. Every bit of directionality in every microphone on the market comes from that gradient term. There is no other way to get it.

The tilt, and the correction that causes the problem

Here is the part that is usually told wrong, including in my own first notes for this page.

Take a sound wave arriving from a long way off, which is effectively a plane wave. The pressure varies sinusoidally in space, so its spatial derivative is proportional to the spatial frequency, which is proportional to the frequency of the sound. So a gradient microphone's raw output rises at 6 decibels per octave: at 1 kilohertz it produces ten times the voltage it produces at 100 hertz for the same sound pressure. Left alone, a directional microphone would be almost unusable, with no bass at all.

So it is not left alone. The capsule is deliberately built so that the diaphragm's motion is mass controlled rather than stiffness controlled across the working band, and a mass controlled diaphragm's velocity falls at 6 decibels per octave for a given driving force. The rising gradient and the falling diaphragm response cancel exactly, and the output comes out flat. That cancellation is fixed in the metalwork. It is not adaptive and it does not know how far away you are.

Now bring the source close. A source nearby is not a plane wave, it is a spherical one, and the pressure of a spherical wave falls as one over the distance as well as oscillating in space. So the radial derivative has two terms, not one: the usual oscillation term, which is proportional to frequency, plus a new term from the amplitude falling off, which is proportional to one over the distance and does not depend on frequency at all. The magnitude of the gradient relative to the far field case comes out as the square root of one plus the square of the speed of sound divided by two pi times frequency times distance.

Read that expression and the whole effect falls out. When the frequency is high or the distance is large, the second term is negligible and the gradient behaves as the far field one did, so the built in correction is right and the response is flat. When the frequency is low and the distance is small, the second term dominates, the gradient is far larger than the fixed correction expects, and the correction overcompensates. The excess appears as boosted bass.

So the proximity effect is not the comparison being frequency dependent. It is a fixed compensation for a far field tilt, being applied to a near field signal that no longer has that tilt. Which also explains, in one line, why an omnidirectional capsule has none: it measures pressure, there is nothing to compensate, so there is nothing to overcompensate. A lavalier can sit an inch from a collar and sound exactly as it would at a foot.

Why the boom is worse than it sounds

Put the numbers next to a voice and the effect is not simply a tonal preference.

A speaking voice puts its fundamental between about 85 and 155 hertz for most men and 165 to 255 for most women. The consonants that carry intelligibility, the sibilants and the stops, live between about 2 and 5 kilohertz. Those two bands are shaded in the drawing, and the point of shading them is that the boost is enormous inside one and nothing inside the other.

At 2 inches a directional microphone adds 20.7 decibels at 100 hertz, 14.8 at 200, 7.5 at 500, 3.3 at 1 kilohertz and 0.3 at 4. At 12 inches those become 6.2, 2.6, 0.5, 0.14 and effectively zero. So moving in by ten inches tilts the voice by 14 decibels across the span from fundamental to consonant, which in power terms is a factor of 25. The voice gets louder and less intelligible in the same movement, which is why close microphone technique sounds impressive on a single word and exhausting over a paragraph.

It also explains the second complaint, the one about sounding raspy as well as boomy. A plosive, the puff of air on a P or a B, is not sound at all. It is a slug of moving air hitting the diaphragm, and a gradient capsule is far more sensitive to it than a pressure capsule, because the moving air produces a pressure difference across the two faces directly. Combine that with 20 decibels of gain at the frequencies where the thump lands and you get the familiar overload on consonants while the rest of the sentence is fine.

Three things that are really going on, kept separate

Three distinct effects get blamed on each other in this category, and the distinction matters because they have three different fixes.

The first is the proximity effect above, which is about distance, affects only directional microphones, and is fixed by moving back.

The second is off axis colouration, and it is specifically a shotgun problem. A shotgun gets its narrow pattern from an interference tube, a slotted tube in front of the capsule. Sound arriving from the side enters at many slots at once with different path lengths and partly cancels, which is the rejection you wanted. But the cancellation depends on wavelength, so sound from the side is not merely quieter, it is tonally mangled, and differently mangled at every angle. That is what makes a shotgun sound strange when a speaker turns their head, and no amount of repositioning the speaker fixes it; you point the microphone instead.

The third is distance itself, acting on noise rather than tone. Sound pressure halves every time you double the distance, which is 6 decibels, and that applies equally to every microphone. So backing a directional microphone off from 2 inches to 12 inches to cure the boom also throws away 15.6 decibels of level. Those 15.6 decibels have to come from somewhere, and where they come from is the gap between your voice and the room.

Which is the trade no listing frames. An omnidirectional capsule has no proximity effect, so you can put it as close as you like, but it also has no rejection, so it hears the room equally in every direction and is only usable because it is close. A directional capsule rejects the room, which is why it can work at a distance, and the closer you bring it the more it colours. The two microphones are solving the same problem from opposite ends, and neither one gives you both.

What our own listings publish, which is almost nothing

Start with the honest end. A Rode VideoMicro compact on camera microphone is 80 millimetres long and 42 grams, needs no battery and draws plug in power from the camera, and ships with a shoe mount, a windscreen and a patch cable. Its record is unusually clear about what it lacks, listing no gain or low cut controls, that it is not as directional as larger shotguns, and that it may pick up handling noise if it is not mounted properly.

The absent low cut matters more than it looks. The standard studio answer to a proximity boost is a high pass filter, and this microphone does not have one. But look again at what a filter could do if it did: a high pass at 100 hertz takes off the left edge of the problem and leaves 7.5 decibels of boost still standing at 500 hertz, inside the vowels, where no filter can touch it without hollowing out the voice entirely. So the missing control is not the limitation it looks like, and the absence of it is honest about where the real fix lives. Also worth saying: the drawback that it is not as directional as larger shotguns is, for this effect, an advantage, because a weaker gradient component means a weaker proximity effect. Full record in our Rode VideoMicro review.

A Rode VideoMic GO II-H shotgun microphone bundle is the other end of the same design, publishing a supercardioid pattern produced by an annular line tube, a self noise figure of 15 decibels A weighted, and dual outputs on 3.5 millimetre and USB-C. Supercardioid means a higher proportion of gradient in the mix, which means a tighter pattern and more rejection, and by the argument above it also means a stronger proximity effect and more off axis colouration. Those are not three features and a flaw. They are one design decision seen four ways. Its record names the self noise as a drawback, which is fair, and that figure is doing separate work in our companion page on bit depth. Details in our Rode VideoMic GO II-H review, and the category is in our shotgun microphones guide, which is right that a tight pattern pushes away room echo and side chatter, and never mentions what it does up close.

Then the lavaliers, where the useful property is the one nobody prints. A BOYA mini 2 wireless lavalier microphone publishes a 5 gram transmitter, 48 kilohertz at 24 bit, adaptive noise cancellation, a limiter and a safety track. A tongyu J3 wireless lavalier microphone publishes a 2.4 gigahertz plug and play link, a noise reduction chip, real time ear monitoring and up to 8 hours of battery. A SYNCO AI mini wireless lavalier microphone publishes noise reduction, voice enhancement and automatic voice levelling over a USB-C receiver.

Not one of the three publishes a polar pattern. Which is the gap, because for a lavalier the pattern is the whole reason it works: an omnidirectional capsule clipped to a collar sounds the same whether the speaker's chin is up or down, whether the clothing has shifted an inch, and whether they turn to the left. A directional capsule at that distance would change tone with every one of those, and no amount of noise processing would hide it. The consistency people like about lavaliers is a property nobody lists. Records are in our BOYA mini 2 review, our tongyu J3 review and our SYNCO AI mini review.

One honest caution about all three, and their own records raise it. Each of them puts a processor in the signal path, variously described as a noise reduction chip, adaptive noise cancellation or automatic voice levelling. Automatic levelling in particular is a gain that moves while you speak, and the artefact is the opposite of a proximity effect: instead of the tone changing with distance, the level changes with content, so the room noise rises in the pauses and drops under the words. The BOYA record is honest that noise cancellation reduces rather than eliminates background noise, the SYNCO record says it is an aid and not magic in extreme noise, and the tongyu record says it is not studio grade. All three are the right caveat. The one nobody adds is that this processing happens before recording and cannot be undone afterwards.

Two of our own pages, one right for the wrong reason and one incomplete

Our microphones for cameras guide says: "Wireless lavaliers, on the other hand, are omnidirectional and pick up sound from all directions, which is great for capturing natural speech when the mic is clipped to clothing."

The fact is correct and the reason is upside down. Picking up sound from all directions is a lavalier's chief disadvantage, not its advantage: it is why a lavalier hears the air conditioning, the room reflections and the person sitting next to you, and why it only works because it is a few inches from the mouth. The real benefit of the omnidirectional pattern is the one the sentence skips, which is tonal stability. No gradient, no proximity effect, so the sound does not change when the geometry does, and the geometry on a clipped microphone changes constantly.

Our guide to getting great audio for video content has the more consequential gap. It says "Get the mic close to the sound source, since that is the whole point of a lav", and it is right, and it is the single most useful piece of advice on the page. But it sits in a section that covers lavaliers, shotguns and desktop microphones together, and the advice is unconditionally right for the first and conditionally wrong for the other two. Get an omnidirectional lavalier close and you gain level with no penalty. Get a supercardioid shotgun or a studio condenser that close and you gain level and 14 decibels of bass tilt. The correction is one clause: get the microphone close, and if it is directional, no closer than about 8 inches unless you want the colouration. Both pages are otherwise sound, and the second is right about the thing that matters most, which is that distance beats equipment.

Problems in our own records

Two worth reporting.

The first is the total absence of the one specification this entire question turns on. Across all nine microphone and recorder products we publish, exactly one, the VideoMic GO II-H, names a polar pattern, and that only because supercardioid appears in its marketing text. None publishes a frequency response plot, a response at more than one distance, or a stated low cut frequency. So a buyer cannot tell from any listing on this site whether a microphone will do this to their voice, and the drawing on this page had to be computed from first principles for an idealised capsule rather than measured from anything we sell.

The second is a factual error on one of our own pages, which is already on our internal defect list for a different reason and is deliberately not linked from here. Our camcorders roundup states that the Rode VideoMic GO II can be powered by a standard 9 volt battery or via plug in power from the camera. Our own product record for that microphone says the opposite and is correct: no batteries needed, bus powered, with no internal recording or battery option listed anywhere. The microphone has no battery compartment. That page is being rebuilt and the claim should go with it.

The honest limit, and the free technique that beats the purchase

The specification that would settle this is a published frequency response at two distances, which is a standard measurement and which no listing in this category gives you. So you cannot buy your way out of the problem by comparing numbers, and there is no filter setting that cures it, because the boost extends up into the vowels where removing it would remove the voice too.

What is left is geometry, and geometry is free. Three moves, in order of how much they buy.

First, distance. For a directional microphone, somewhere around 8 to 12 inches puts the boost at a couple of decibels, which is a pleasant warmth rather than a problem. Inside about 4 inches it stops being a tonal choice. If you want the broadcast sound on purpose, that is a legitimate use of the effect and radio presenters have used it for eighty years, but choose it rather than arriving at it.

Second, angle. A gradient microphone responds to the component of the gradient along its own axis, so speaking slightly across the capsule rather than straight into it reduces the gradient term without changing your distance. Thirty degrees off axis costs you very little level on a cardioid and moves the plosive blast past the diaphragm instead of into it. This is what a radio presenter is doing when they talk past the microphone rather than at it, and it is the single most effective thing on this list.

Third, and this is the one that resolves the whole trade: if you want to be close, use an omnidirectional capsule, which is what a lavalier is. A lavalier at 6 inches gets 6.3 decibels more level than a directional microphone at 12 inches and adds no tilt at all, because it has no gradient term to over correct. That is the entire reason the category exists and it is a better reason than any of the ones our own guides give for it.

And the arithmetic to carry away, because it connects to the other half of this argument. Moving a microphone closer by one halving of distance gains 6 decibels, and 6 decibels is exactly one bit of dynamic range. On a directional microphone that bit arrives with a 7 decibel bass tilt attached. On an omnidirectional one it arrives clean. We work the rest of that currency out in why 24 bit matters and 96 kHz does not, where moving 18 inches closer turns out to be worth two bits and the step from 16 bit to 24 bit turns out to be worth mostly nothing.