On a phone the camera is rarely the problem and the microphone almost always is. Getting the mic within a foot of the speaker, choosing a small soft room over a large hard one, locking focus and exposure, and switching on aeroplane mode will improve a recording more than any purchase under a few hundred pounds.
Flow diagram of 5 steps: Record thirty seconds of the empty room and actually listen to it, Switch off everything switchable, and write the list down, Put every phone in the room into aeroplane mode with notifications silenced, Check storage and battery, then plug in, Lock focus and exposure on the subject’s face.
A modern phone camera is a genuinely capable instrument for an interview. The microphone a few millimetres away from it is not, and the gap between those two things is the most useful fact to hold on to before shooting anything. An audience will watch soft, slightly noisy pictures for twenty minutes without complaining. They will not sit through twenty minutes of a voice arriving from the far side of a hard room, and most of them will leave inside the first thirty seconds without being able to say what was wrong.
The reason is not that phone microphones are cheap components. They are small, but they are decent. The problem is where they sit and what they were designed to do. A phone carries two or three microphones placed for the person holding it: one at the bottom edge, near the mouth of somebody making a call, and one at the top for noise cancellation. The processing built on top assumes the speaker is the phone holder, roughly a hand’s width away, probably on a windy street. Set that phone on a coffee table two metres from an interviewee in a living room and every assumption in the chain is now wrong, which is why the result sounds hollow rather than merely quiet.
Distance is the mechanism, and it behaves in a way that is worth understanding rather than memorising. Sound pressure from a source falls away steeply with distance, so closing the gap between mouth and microphone gains you a large amount of the voice you actually want. The reflected sound bouncing off walls, ceiling and tabletop does not fall away in the same manner, because it is arriving from every direction at once. What changes as the microphone moves closer is therefore not simply level but the ratio of direct voice to room, and it is that ratio, rather than the volume, that an ear reads as competent or amateur.
This is also why the problem cannot be deferred. Volume is trivial to fix in an edit. Ratio is not: once a room is baked into a recording there is no plugin that lifts it out without taking the texture of the person with it. Noise-reduction tools have improved considerably and they are still a way of choosing which damage you prefer. The thirty seconds spent moving a phone closer before the first question is worth more than any amount of processing afterwards.
With nothing bought at all, that yields one instruction: get the microphone close, and keep it still. A phone propped against a stack of books at chest height, thirty to forty centimetres from the speaker and angled slightly to one side of their mouth, will out-record the same phone on a tripod across the room by a margin nobody will need pointing out. Slightly to one side matters, because it keeps plosives — the small blast of air on p and b sounds — from striking the microphone head-on.
What not to do is hold it. A hand on the case transmits every small adjustment of the fingers into the microphone as a low thump, and the phone’s own noise processing tends to duck the voice when it hears one, so a rustle costs you a word as well as adding a knock. Anything solid beats a hand: a mug, a folded jumper, a bag of rice, a five-pound clamp from a hardware shop. If it is going to sit there for an hour, tape it down.
The room contributes more to a recording than any single piece of equipment, and unlike equipment it is free to choose. What you want is a small, soft, cluttered space. What you are avoiding is a large, hard, empty one.
Soft means carpet, curtains, upholstery, a full bookshelf, a bed, coats on hooks. Each of those absorbs the mid and high frequencies that make speech intelligible, instead of returning them to the microphone a few milliseconds late as a smear. Hard parallel surfaces do the opposite. Two bare walls facing each other across three metres produce a flutter that is close to inaudible while you are standing in the room and completely unmistakable on playback, because your ears do a job in the room that a microphone will not do for you.
The rooms that sound worst tend to be the ones that look most convenient. Kitchens have tiles, a fridge, an extractor hood and hard cabinet doors. Bathrooms are the same problem in miniature. Offices and meeting rooms are built from plasterboard, glass and hard flooring, and they usually contain ventilation running on a cycle you will only discover in the edit. Church halls, atria, school gyms and any room with a high ceiling are unusable for speech at any distance past about a foot. A bedroom with the curtains drawn is, unglamorously, close to the best free option most people have access to.
Then there is noise, where the distinction that matters is between noise you control and noise you do not. Fridges, freezers, boilers, extractor fans, air conditioning, ticking clocks, laptop fans, aquarium pumps and buzzing standby transformers are all switchable, and switching them off costs nothing but memory. Write down what you turned off, because a freezer left unplugged after a shoot is a real way to ruin somebody’s week. Aircraft, traffic, neighbours, plumbing in other flats and weather are not switchable, so the question there becomes whether the noise is constant or intermittent. Constant noise is survivable — the ear adapts to it and an editor can work around a steady floor. Intermittent noise is what forces retakes, because a lorry passing through the middle of the best answer of the day cannot be lifted out from under the voice.
None of this requires a survey. It requires listening to the room on headphones for thirty seconds before anyone sits down, which is the step that almost everyone skips and the one that catches nearly everything.
Press record, stay silent, and play it back on wired headphones at a normal listening level. This is the only reliable way to hear what the microphone hears rather than what your ears have already filtered out. Hum, hiss, a distant fan and a ringing quality in the reverb all show up immediately. If the empty room sounds bad, no interview recorded in it will sound good.
Fridge, freezer, extractor, boiler, air conditioning, fans, radios, laptops, anything with a cooling cycle. Keep the list on your phone rather than in your head. Restoring power to somebody’s appliances at the end of a shoot is part of the job, and a forgotten freezer is a much more expensive mistake than a bad take.
On many phones an incoming call will stop a recording outright, and there is no warning and no partial file worth having. Notifications arrive as an audible tick in the audio even when the ringer is off, because the haptic motor is bolted to the same chassis as the microphone. This applies to the interviewee’s phone in their pocket as much as to yours.
Work out roughly how many minutes you intend to record, double it, and confirm the space exists before anyone arrives. Run from mains power if there is a socket. A recording that stops at forty minutes because the disk filled is unrecoverable in a way that almost nothing else on this list is.
Frame the shot, then press and hold on the face until the lock indicator appears, and leave it there. Autofocus hunting and exposure shifting every time somebody moves a hand are the two clearest signs of an unmanaged phone. Re-lock if the framing or the light changes, and check it again after any break.
Say the date, the name of the interviewee and which take it is, straight into the camera, then clap once with your hands in shot. The slate saves an editor from guessing at forty unlabelled files. The clap produces a sharp spike that lines the second recorder up against the picture in seconds.
Ask everyone to stay still and silent, and record the room as it was during the interview. This is the material an editor uses to patch the gaps left by cuts, so that removing a cough does not leave a hole of pure silence that the audience hears as an edit. It takes half a minute and it is missed constantly.
Most of the settings arguments people have about phone video are irrelevant to an interview, which is a static shot of a person talking in a room. A handful of choices genuinely change whether the material is usable, and the rest are preference. The table covers the ones worth setting deliberately before the first question rather than discovering afterwards.
Phone settings for a sit-down interview, and the reason behind each
| Setting | What to choose | Why it matters |
|---|---|---|
| Frame rate | 25fps in the UK and Europe, 30fps in North America | Matching the local mains frequency keeps LED and fluorescent lighting from flickering or rolling banding through the frame. It is the one setting that can render footage unusable outright. |
| Resolution | 1080p unless you have a reason for 4K | Roughly 60 MB a minute at 1080p30 on recent iPhones and around three times that at 4K, with Android varying by codec. 4K buys crop latitude and reframing room; it costs storage, battery and heat. |
| Focus and exposure | Locked manually on the face, not automatic | Autofocus hunting mid-answer and exposure pumping as somebody gestures are the most recognisable amateur tells, and neither is fixable later. |
| Lens | The main wide lens, never digital zoom | The ultrawide distorts faces at conversational distance and the telephoto has a smaller sensor that struggles indoors. Digital zoom is a crop, so move the phone instead of zooming. |
| Aeroplane mode | On, with notifications silenced | A call can end a recording with no usable file, and a silenced notification still reaches the microphone through the haptic motor as a tick. |
| Stabilisation | On handheld, off on a fixed mount | Stabilisation crops the frame and can introduce a slight warping. On a locked-off tripod it has nothing useful to do and is quietly costing you image area. |
| Orientation | Horizontal for anything destined for a 16:9 player or a television | Shooting vertical and cropping to 16:9 later discards most of the frame. If a vertical cut is also wanted, shoot 4K horizontal and crop a vertical version out of it. |
If two phones are available — yours and the interviewee’s, or yours and a borrowed one — the largest single improvement available at no cost is to stop asking one device to do both jobs. One phone films from wherever the shot looks right. The second sits close to the speaker doing nothing but recording sound. This decouples the two constraints that were fighting each other, because the camera wants to be far enough away to frame a person and the microphone wants to be close enough to beat the room.
Put the recording phone somewhere solid, thirty to forty centimetres from the speaker and out of the camera’s frame — behind a laptop, on a cushion just below the bottom of shot, clipped to a chair back. Use a plain voice-recorder app rather than the camera app, because the camera app on many phones applies aggressive voice isolation and wind suppression that produce audible artefacts on a close, quiet source. If the phone offers an uncompressed or high-bitrate option, take it. Then check the level by recording a real answer, not a test count, because people speak more quietly when they are being interviewed than when they are demonstrating a microphone.
Phones have no shared clock you can rely on, so the sync happens through a transient you deliberately create. Clap once, sharply, with your hands inside the camera frame, at the start of every take. In the edit, both files show the clap as a narrow spike; line the spikes up and the two recordings run together. Most editors will also do this automatically by analysing the audio, using the camera phone’s own poor-quality track as the reference — which is a reason to keep that track running rather than muting the camera. Expect a slow drift on takes past about twenty minutes, and split long interviews into shorter takes to avoid chasing it.
A clip-on lavalier microphone is the cheapest real upgrade there is, because it fixes the distance problem permanently: the microphone is now a fixed few inches from the mouth and it stays there whatever the subject does. The trap is the connector. A 3.5mm lavalier is usually wired TRS for a camera, and a phone headset socket expects TRRS, so the two are not interchangeable and the wrong combination produces silence or a faint mono channel — and most modern phones have no socket at all, so you are looking for a USB-C or Lightning version, or an adapter that carries an audio interface rather than a passive splitter. Check the connector against the actual phone before buying anything. Money spent on cheap wireless before you own a reliable wired lav is money spent on new failure modes: dropouts, interference in crowded spectrum, and batteries that die mid-answer.
One microphone cannot serve two people sitting a metre and a half apart. Placed between them it is too far from both and picks up the room instead; placed near one it makes the other sound like they are in the next room, and the imbalance is worse than a consistently mediocre recording because the audience notices it every time the conversation turns. The options are two lavaliers into a small interface, two phones each recording one person, or a genuinely close single microphone that you hand between speakers and accept the handling noise from. A phone on the table between two chairs is the arrangement to avoid, and it is the one people default to.
Wear headphones and listen to what is being recorded while it is being recorded. Every problem on this page is obvious through headphones and invisible without them, and the cost of not knowing is a second interview with somebody who has already told you the story once. Use wired headphones: Bluetooth introduces enough latency to be distracting, and on some phones connecting a Bluetooth headset silently switches the input to the headset’s own microphone, which is how people record an hour of their own ear. If the recording device cannot be monitored while it records, listen back to the first two minutes before continuing.
A phone is not a small broadcast camera, and pretending otherwise wastes a shooting day. There are specific conditions where the honest answer is to change the plan rather than to try harder.
Low light is the first and least negotiable. A phone sensor is physically tiny, and in a dim room it compensates with noise reduction and sharpening that turn skin into something smooth and slightly plastic, with detail smeared around the eyes and hairline. It does not look like grainy film; it looks like a processed image, and no amount of grading recovers it. The fix is light rather than settings: put the subject beside a window with the window behind the camera, or add one soft lamp bounced off a white wall or ceiling. A single cheap lamp bounced off something white does more for a phone image than a more expensive phone would.
Long unbroken takes are the second. Recording video is one of the hardest things a phone does, and sustained recording heats the processor until thermal management intervenes. On a warm day, in sunlight, or with the screen at full brightness, a phone can stop recording partway through without a usable file. Interviews longer than about twenty minutes should be broken into separate takes at natural pauses, which also happens to make the sync drift and the storage arithmetic easier.
Reverberant and public spaces are the third. A café, a foyer, a market, a gallery or a hall will defeat a phone microphone at any distance that also produces a reasonable shot. If the location is the point of the piece, record the interview somewhere controllable and shoot the location separately as material to cut to, rather than trying to do both at once and getting neither.
The fourth is anything requiring shallow depth of field or genuine optical reach. Portrait modes estimate depth and guess wrongly at the edges of hair and glasses, which is distracting in motion in a way it is not in a still photograph. If a shot needs a background to fall away, create the separation physically: move the subject further from the wall behind them and closer to the lens, rather than asking software to simulate it.
It is worth being straight about this platform too, because the surrounding advice on a lot of sites quietly implies a support structure. Kind Channel is new. Nothing has aired, the community is small, and there is no equipment loan scheme, no technical desk that will check your levels before you shoot, and no post-production team that will rescue a recording made in a tiled kitchen. Everything above is what an individual can do alone, with a phone they already own and possibly one borrowed second phone, which is the actual situation most people proposing a first interview are in. A proposal is easier to assess when it says plainly where the interview will be recorded and what it will be recorded on, because those two answers predict most of what the finished piece will sound like.
No, but you do need to solve the problem a microphone solves, which is distance. A phone’s built-in microphones are positioned and processed for somebody holding the phone near their face, so they perform badly across a room. Propping the phone thirty to forty centimetres from the speaker, slightly off to one side of their mouth, on something solid rather than in a hand, gets most of the benefit for nothing. If you do buy one item, a wired clip-on lavalier is the cheapest thing that changes the outcome permanently — check its connector against your specific phone first, since a 3.5mm lav wired for a camera will not work in a phone headset socket.
Decide by where the finished piece will play rather than by how the phone feels in the hand. Anything destined for a television, a 16:9 player or a broadcast schedule should be shot horizontally, because cropping a vertical recording to 16:9 later throws away most of the frame and leaves you with a soft, tightly framed image. If both a horizontal and a vertical version are needed, shoot horizontally at 4K and crop the vertical cut out of the larger frame. Shooting vertical is defensible only when vertical is the sole destination.
Because the sensor is physically small, and in low light the phone compensates with heavy noise reduction and sharpening. That processing smooths skin into something slightly plastic and smears fine detail around eyes and hairlines. It is not recoverable in grading and it is not a settings problem. The answer is more light: seat the subject beside a window with the window behind the camera, or bounce a single soft lamp off a white wall or ceiling. One cheap lamp used well improves a phone image more than upgrading the phone would.
Avoid it. Bluetooth audio is heavily compressed for voice calls, so the recording arrives thin and artefact-laden before the room even gets involved. Latency also makes monitoring through the same earbuds distracting, and on some phones connecting a Bluetooth headset silently switches the recording input to the headset microphone, which is how people end up with an hour of audio recorded from beside their own ear. Use wired headphones for monitoring and either the phone’s own microphone placed close, a wired lavalier, or a second phone as a dedicated recorder.