Microphone proximity effect

Why does a singer's voice gain weight and bass simply by moving closer to the microphone?

A singer standing a foot from the microphone sounds ordinary. They lean in until their lips almost touch it, sing no differently, and the voice arrives fuller and warmer, with an obvious weight at the bottom that was not there a second earlier.

Moving closer should make everything louder together, and it does. But that is not what happened: the low frequencies came up far more than the high ones. Something is treating bass and treble differently as a function of distance alone. What, and why only for some microphones?

Ask what a microphone is measuring, because there are two quite different answers.

One kind seals the back of its diaphragm inside a closed capsule, so the diaphragm feels the absolute air pressure at one point. Pressure at a point has no direction, so it hears equally from all sides — an omnidirectional, or pressure, microphone.

The other leaves both faces open to the air, so it is driven not by the pressure at a point but by the difference between its front and back faces, a couple of millimetres apart. That is a pressure-gradient microphone, and its directionality falls out of the design: sound arriving edge-on reaches both faces at the same instant with the same strength, so the difference is zero and the microphone is deaf to it. The figure-of-eight pattern is not added on; it is what measuring a difference means.

So what makes the pressure at two nearby points differ? Exactly two contributions, behaving completely differently.

The first is timing. Sound from the front reaches the back face slightly later, so the faces sit at different points in the same oscillation, by an amount that depends on how much of a cycle the gap represents — which grows with frequency. At low frequencies the wavelength dwarfs the gap, the faces are nearly in step, and this contribution nearly vanishes.

The second is amplitude. Sound from a point source thins as one over the distance, and the front face is at r while the back is at r plus the gap. At three metres, a two-millimetre gap changes the distance by less than a tenth of a percent, so the difference is negligible; at three centimetres it is a serious fraction. This contribution has nothing to do with frequency — only with how steeply the field is thinning where the diaphragm sits.

Put them together and the effect falls out. Far away only the timing term survives, and because it rises with frequency the microphone is equalised to cancel that rise — which is what makes a gradient microphone flat at working distance. Move in close and the amplitude term arrives, unequalised and frequency-independent, dominating precisely at the low frequencies where the timing term had almost nothing to offer.

The arithmetic is clean enough to do here. For an ideal point source, a pure gradient microphone's response relative to its far-field value is ten times the base-ten logarithm of one plus one over kr squared, where r is the distance and k is two pi times frequency over the speed of sound. At 100 Hz, with sound at 343 metres per second, k is about 1.83 per metre. At one metre that gives about 1.1 dB of lift; at 30 cm, 6.3 dB; at 15 cm, 11.5 dB; at 5 cm, 20.8 dB. The same 5 cm figure at 1 kHz gives only about 3.4 dB. So a singer at 5 cm gets roughly 17 dB more help at 100 Hz than at 1 kHz — audible as "weight" rather than as loudness.

The misconception first: the singer's voice is not gaining bass. Nothing about the source changed. A frequency-dependent gain has been applied by the transducer, and a matching low-cut removes it again — which is why engineers treat working distance as a tone control.

Second, most vocal microphones are neither pure pressure nor pure gradient. A cardioid is a blend, so it shows a real but smaller effect than the figures above, which assume a pure gradient and an idealised point source — and a mouth a few centimetres away is not a point source. Treat the numbers as the shape of the phenomenon, not a specification.

Third — and here the perceptual and the conventional part company — the bass lift is physics, but the taste for it is not. Close working also raises direct sound relative to the room's reflections, a separate mechanism supplying much of what we call intimacy; Concert hall reverberation covers that side. And the association of a warm close-miked voice with authority and confiding is learned from decades of broadcast and crooned singing, possible only once amplification let a singer stop projecting. There is no evidence a low-tilted voice is intrinsically trustworthy; there is very good evidence we have all heard thousands of them.

Keep this apart from Voice recording mismatch: there the low-frequency reinforcement is added inside the listener's own skull and is missing from every recording, while here it is added by the microphone and is present in the recording everyone else hears.

How would we know the transducer is responsible rather than the source or the room? Put an omnidirectional and a figure-of-eight side by side on the same source, at one metre and at five centimetres. This account predicts a large low-frequency tilt on the gradient microphone and essentially none on the omnidirectional. The observation that would sink it: the omnidirectional showing the same close-up bass rise, meaning the effect belongs to the sound field or the singer rather than to the mounting of the diaphragm.

• Why an omnidirectional microphone is the choice when a close voice must not be coloured.
• Whether the taste for a weighted close voice is stable across musical cultures, or specific to those raised on broadcast.

Every term the collection defines is gathered in the glossary.