Recording Rain in a Bucket

Roughly half our sound library was neither bought nor captured in the wild. It was built indoors, at unsociable hours, out of household objects. This is an account of how — not a tutorial: our equipment is mid-range and several of our recordings are poor, and we will name those too.

Five nights from the diary

We keep a field notebook, mostly to stop repeating mistakes. Five entries, lightly tidied.

Night one — the good roof. Steady rain, corrugated roof, nobody around. Perfect conditions, unusable recording: too much high frequency, no low end, a constant level with no shape. Real rain under a metal roof is mostly noise, and noise is what the brain throws away.

Night two — the wrong success. A drainpipe with a beautiful irregular drip. Forty delighted minutes of it, unusable for the opposite reason: the drip has too much character. A recognisable event that recurs is the fastest way to make somebody notice they are listening to a file. That was the beginning of a rule.

Night three — a car. Two hours of excellent courtyard rain, ruined at minute fifty by a car alarm and again at seventy by a car door, both in the only good stretch. City recording is mostly the discovery that nothing is ever quiet enough.

Night four — four in the morning. The only reliably quiet hour in a city runs from about half three to half five, and even then one lorry takes the session. We plan around this now rather than hoping. It is much of why our library is made indoors: a kitchen at eight in the evening is quieter than a courtyard at four.

Night five — the balcony and the bucket. A tin bucket half full of water, a square of muslin over half of it, a watering can with a fine rose poured from about a metre. The muslin softens the impacts into something diffuse; the open half gives you the occasional single drop. Ten minutes of that under a soft wide bed is the rain in most of our episodes.

The real rain sounded like static. The bucket sounded like rain. We have stopped being surprised by this and started planning for it.

Realistic and believable are different words

This is the idea underneath every other decision here, and it took an embarrassingly long time to state.

A microphone records everything that reaches it, equally. Ears do not — they discard, sharpen, and attend to some things and not others. So a faithful recording of a place does not sound like being there; it sounds like a recording of it. The hiss, the traffic floor, the frequencies attention would have thrown away, all preserved as carefully as the parts you would actually have noticed.

Foley done well is not fake. It is a reconstruction of what a person would have heard rather than what was there, and that is why a bucket beats a rainstorm.

The kitchen is the studio

Most of the close, warm sounds in our episodes were made within three metres of a kettle. None of this needs good equipment; it needs a quiet hour.

1. 1

   Footsteps on a wooden floor

   A pair of shoes worn on the hands, on a board of the right wood, beats walking. You get the weight where you want it and none of the body noise or clothing rustle that ruins a real take. Wood on wood, close, at half the volume you expect.

2. 2

   Fabric and clothing

   A wool jumper moved slowly near the microphone, never at the speed of the action on screen — slower reads as heavier. Every blanket in the catalogue is the same jumper, used in about thirty episodes.

3. 3

   Vessels — the easy win

   Cups, pots, spoons, the hollow tap of a ceramic lid replaced slightly off-centre. Real objects, close, deliberately imperfect. No trickery needed: a real cup on a real saucer sounds exactly like itself, which is not true of most things.

4. 4

   Doors, and the trick of the latch

   Nobody hears a door; everybody hears a latch. We record the small metal part and let the swing stay almost silent, and it reads as a door far more clearly than a recording of a door does.

5. 5

   Fire, which we still cannot do

   Cellophane is the traditional answer and it sounds like cellophane to anyone paying attention. Ours mixes a badly recorded real fire with small crackles placed by hand, and it is the weakest part of the library. We keep it low in the mix, which is a polite way of saying we hide it.

Not all of it is homemade, and it is worth saying which part is not. We buy about a fifth of what ends up in an episode.

Anything needing a place we cannot reach: a particular forest, a harbour, distant church bells, cicadas of the right species. Anything needing equipment we do not own — genuinely low, quiet rumbles want microphones several times more expensive than ours. And occasionally a bed layer better than anything we managed, bought because the schedule ran out rather than because the recording was impossible.

We license these properly, and we mention it for one reason: a workshop piece about making your own sounds that quietly leans on purchased ones is a slightly dishonest piece. The bucket story is true and it is not the whole library.

The most boring job

Labelling: the least interesting work in the process, and the part that decides whether any of the above was worth doing.

A recording that cannot be found does not exist. We learned this by remaking a rain layer we already had, twice, because it lived in a folder called new inside a folder called rain 2. Every file now carries its material, distance, length, whether it loops cleanly, and — the field that matters most — whether it has character. That last one is a yes or no, and it decides whether the sound can ever go in a bed layer.

Labelling costs about a fifth of the recording time and nobody enjoys it. Skip it once and the session was a hobby rather than a library.

Where our sound is weak

The honest list.

The first eight-hour rain window. The rain bed is thin and harsh in the high frequencies — an early attempt with too much real recording in it. Fine on a phone speaker, tiring on good headphones after twenty minutes, and it is our most-played file.

The winter-hearth long version. An identifiable creak in the nine-minute room-tone bed — a chair, in our own recording room, which we did not hear until several viewers did. It breaks the rule from night two, after we wrote the rule down.

The first harvest-day episode. The wind is a purchased layer that does not match the trees: a wide open-field wind under a shot of dense woodland. Nobody has mentioned it. It is the one that makes us wince.

Every fire we have recorded. Still hiding it, as above.

Why the bucket wins

Not because it is clever, and not because we could not afford the rain. It wins because a bucket is the shape of what a person remembers, while a microphone in a real storm is the shape of what was there — two different objects.

The listener's side of this is over in the sound of rain. This side is a balcony at four in the morning, a watering can, and a neighbour who has never asked what we are doing and by now presumably has a theory.

The evening is warm here.