I spent part of this week shopping for rain.
Not literally. I've been building a long spoken-word piece for my own listening, the kind of thing meant to play at night with the lights off, and it needs a bed under the voice. Something steady. Something that runs for twenty or thirty minutes without a car door or a dog bark in the middle of it. Rain, or the ocean.
I went in assuming the choice was about mood. The ocean is slower and bigger. Rain is closer and busier. Pick the one that feels right, set the level, done.
That's not what happened. The choice turned out to be about frequency, and the recording I liked best was the one that would have buried the voice.
Looking at it before listening to it
There are thousands of long nature recordings out there, and auditioning half-hour files one at a time is a good way to lose a week. So I screened them the lazy way first. Most of the big sound libraries show a waveform and a spectrogram for every file, and those pictures say a lot before a single second plays.
The waveform tells me whether the file is even. A long, flat band means steady. A spike means thunder, or a wave slamming a rock, or somebody's zipper. I threw out anything with spikes.
The spectrogram tells me where the energy lives. And that's where the ocean lost.
Surf sits right where the voice sits
Rain on a roof is mostly high. The hiss and patter live up in the top end, a few thousand hertz and above, with not much underneath. It's a bright, airy texture.
Surf is the opposite. The big rolling energy of breaking waves piles up in the low mids, roughly 200 to 500 Hz. On the spectrograms it showed up as a thick band in exactly that range, recording after recording.
The voice also lives in that range. The body of a speaking voice, the warmth that makes it sound like a person in the room instead of a phone call, is low mids. Put a thick wall of surf under it at any real level and the two fight for the same space. The voice doesn't get quieter. It gets blurry. The consonants still come through, but the chest goes, and what's left sounds like someone talking from the next room.
So the recording I'd have picked by vibe was the one that would have made the voice worse. The rain, which I liked less on its own, leaves the low mids almost empty. It wraps around the voice instead of sitting on top of it.
If I do end up using an ocean bed, it gets an EQ dip in that low-mid band before it goes anywhere near the voice. That's the plan on paper, anyway. I haven't printed it yet.
This is a song problem too
None of this is new information. Every mixing engineer knows that two things in the same frequency range compete. But I've been treating beds as atmosphere for years, and atmosphere gets a pass. I pick a pad or a room tone or a drone by how it feels, and then I wonder why the vocal won't sit.
Going back through a few of my own sessions with this in mind, the pattern was obvious:
The warm pad under the verse. It sounds gorgeous soloed. It's also a fat block of 250 to 400 Hz sitting exactly where the vocal's body is. Every time I've pushed the vocal up to get it out of that pad, I've made the vocal harsh instead of clear, because the top end got louder and the body was still buried.
The ambient guitar swell. A volume-swelled chord through a big reverb is one of my favorite sounds. It's also mostly low mids once the reverb tail fills in. Under a sung verse, it's surf.
The room mic on the acoustic. Lovely on its own. Boxy under a vocal.
In every case the bed itself was fine. It was the wrong shape for what was sitting on it.
Picking the bed by where it leaves room
What changed for me this week is the order of questions. I used to ask "does this sound good?" and then "does it fit?" Now I ask the second one first, and I ask it in a specific way: where is this thing going to leave room for the voice?
A bed that lives above the voice, like rain, like a high shimmering pad, like airy cymbal wash, can sit at a surprising level and still let the vocal through. A bed that lives below it, a sub drone or a low string note, mostly stays out of the way too. The danger zone is the band the voice actually needs, and that's where most of the "warm" textures I reach for happen to live.
The honest test is simple. Solo the voice and the bed together, nothing else, and listen to the voice only. If it still sounds like the same person in the same room, the bed is fine. If it sounds like they stepped back three feet, the bed is in the way, and no amount of vocal level is going to fix it.
For the late-night piece, it's rain. I like the ocean more. The voice likes the rain more, and the voice is the point.