Lavalier microphone mistakes

A clip microphone is the most reliable sound upgrade available to somebody filming alone. It also produces footage nobody can rescue. Both are true for the same reason: it is attached to a person, and people move.

Start with the centre line, keep moving fabric clear and leave a small loop of slack in the cable. Then record a test with the movements the person will make. A setup that works while standing still still needs that second check.

Where does a lavalier go?

A lavalier, the small clip microphone that goes on your chest, wants to sit roughly a hand's width below your chin, on your centre line, attached to something that will not move independently of you.

Centre line matters more than people expect. Clip it to the lapel on your left and every time you turn right your voice arrives quieter and duller, because your own head is now partly between your mouth and the microphone. Talking to one camera and never turning, an off-centre position survives. Looking between two cameras, or down at a hob and back up, it does not: the level moves every time and the edit inherits a voice that fades in and out.

Too low gives a distant, chesty sound with clothing near it. Too high catches breathing and swallowing. A hand's width is a starting point, overruled like every starting point in audio by what you hear on playback.

Keeping the capsule clear

The capsule is the small part at the end that hears things. Anything brushing it is louder than everything you say.

Hair first. Long hair moves whenever the head moves, drags across the microphone with a noise like static, and the person wearing it is the one person in the room who cannot hear it. Tie it back, or clip the microphone where the hair does not reach, and check by turning your head both ways with the recording running.

Jewellery next. A chain resting on a microphone ticks against it all session, and rings count if you talk with your hands near your chest, which cooks and instrument teachers do constantly. Then the parts of your clothing that move on their own: a loose collar, a scarf, a lanyard, a swinging jacket.

Which fabrics rustle worst, and why?

Stiff, slippery and synthetic. Waterproofs are the worst thing to wear near a clip microphone: technical jackets, shell fabrics, anything that makes a noise when you move your arm in a quiet room. Starched shirts and stiff linen come next, and silk is worse than it looks. Soft, matte and heavy is the safe end: cotton jersey, soft knitwear, anything with enough weight to hang still.

The mechanism tells you what to do when you cannot change clothes. The noise is not the fabric. It is fabric moving across the capsule, so the fix is to stop the movement rather than change the material. Clip through the fabric so the microphone is held still. Tape the cable inside the garment above and below the clip, so the whole assembly travels with the body. If the microphone goes under clothing, tape the material into a small tent above it so nothing touches the capsule.

Why does the cable need a loop in it?

Because the cable is a route into the microphone for everything happening further down it.

Leave a small loop of slack just below the clip and secure it so it holds its shape. That loop is strain relief: when the cable is tugged, or an arm catches it, or the recorder in your pocket shifts, the loop takes up the movement instead of passing it into the capsule.

Without one, every tug lands as a thump, and the thumps arrive on the moments where you moved, generally the moment you were demonstrating something.

Why is rustle difficult to remove?

Because it lives in the same place as your voice.

Steady noise gives a profile-based tool something consistent to estimate. Even then, reduction can affect the wanted sound. A changing rustle is harder to distinguish from speech, so prevention leaves the repair fewer compromises to make.

A rustle has no shape. It is a short, broadband burst, meaning it covers a wide spread of frequencies at once, and that spread largely carries the consonants in speech. Subtract the rustle and the consonants go with it, so words lose their edges and the voice comes back slurred. Leave it in and the rustle is audible. That was the whole story until recently; the newer machine-learning de-rustle tools are the third option, trained to tell rustle from speech, and they lift a good part of it, though not perfectly and not reliably enough to lean on. Which is why a rustle recorded is still, in practice, mostly a rustle delivered, and the fix worth having is the thirty seconds before you press record.

These checks protect the recording before a repair becomes necessary. They cannot guarantee a fault-free take, so listen again after a break or a change of clothing.

The list you write once and use every time

Hair tied back or clipped clear. Jewellery off or secured. Nothing stiff or waterproof over the capsule. Microphone on the centre line, a hand's width below the chin. Cable looped, secured, running down inside the clothing. One sentence recorded and played back with the head turned both ways before the real take begins.

That last item takes thirty seconds and catches most of the ones above it. It feels like a formality until the first time it saves an afternoon.

This is worth handing over before a shoot rather than discussing after one. A cooking partner working with this studio was sent a written pre-shoot list before filming, and hair touching the lavalier was on it, alongside what to check when resuming after a break and which angles would be wanted later. None of it is difficult, and all of it is invisible until it is missing.

Keep your version where you get dressed. A checklist that lives in the edit suite is a record of what went wrong.

Sources and evidence

What here is established audio practice and what is our own method.

What is established. Where a small microphone sits, and how its tone shifts as the body turns toward or away from it, follows from the proximity and off-axis behaviour documented by DPA Microphones. Why a clothing rustle resists removal is a matter of signal rather than effort: it is a broadband burst occupying the same frequencies as the consonants in speech, so subtracting it takes the consonants with it, the trade set out in peer-reviewed speech-enhancement research; the newer machine-learning de-rustle tools narrow that trade without closing it.

Our method. Preventing the rustle at the clip, and handing a partner a written pre-shoot checklist rather than discussing it afterwards, is the practice here.

References.

• DPA Microphones. Proximity effect in microphones explained. dpamicrophones.com. Distance response depends on microphone design and polar pattern. Manufacturer education.
• Kim, G. and Loizou, P. C. Gain-induced speech distortions and the absence of intelligibility benefit with existing noise-reduction algorithms. pmc.ncbi.nlm.nih.gov. Noise suppression introduces amplification/attenuation distortion; conventional NR does not by itself raise intelligibility. Constraining amplification distortion CAN yield gains; does not establish a viewer-retention effect.
• Multiple authors. Speech enhancement artefacts and naturalness. arxiv.org. Modern enhancement can introduce artefacts; naturalness remains a constraint. Preprint; use as supporting rather than sole evidence.