What makes rain sound like rain?
Drops. A shower and a rainstorm have a similar spread of frequencies, but only rain is made of separate impacts: a short tick each time a drop hits a surface, and now and then a small rising "plink" when one lands in standing water. Thousands of distant drops blur into a steady hiss. The handful close enough to hear one by one are what tell your ear it is raining.
That is why this player does not simply filter noise until it sounds wet. Filtered noise at rain's spectrum reads as a shower or as static, because the individual drops are missing. Here every drop is synthesized as its own event and laid over a shaped bed of noise. If you want the softness of rain with no drops at all, pink noise is the closest steady sound.
Light rain or heavy rain: which suits sleep?
The two rains differ in how many drops fall, how bright they are, and how much low end sits underneath them. Those numbers are listed here so you can see why they sound the way they do.
| Rain | Character | Drops | Low end | Closest to |
|---|---|---|---|---|
| Light Rain | Sparse, thin, gentle | 13 a second | Rolled off below about 800 Hz | Rain on a window from inside a quiet room. |
| Heavy Rain | Dense, full, steady | 85 a second | Kept down to about 190 Hz | A downpour on the roof and the street outside. |
Light rain leaves space between drops, so it sits quietly in the background and is easy to listen to as you drift off. That space is also its limit: a voice or a car can slip through the gaps. Heavy rain is dense enough that the drops fuse into one continuous sound, and its low end covers rumble from a road or a boiler, so it is the better choice for a noisy room. Switching crossfades over a third of a second, so you can compare the two without a gap.
Why synthesized rain beats a looped recording.
A rain recording is one storm, in one place, on one day. It carries that room's reflections, the gutter, the car that went past, and none of it can be removed afterwards. It also has to loop, and rain is one of the hardest sounds to loop well: your ear remembers events, so one loud drop or gust coming back on a schedule becomes the thing you wait for.
Nothing here is a clip. Each drop is placed at random, with its own loudness, pitch and length, and the drops are stamped into two long layers of different lengths, each with its own rain. The layers themselves are finite, but because their lengths share no common factor, the combined pattern does not line up again for about 283 days at the 22.05 kHz rate this player runs at, and longer at higher rates. There is no thunderclap to wake you and no room baked in.
How this generator actually works.
Each layer starts as white noise shaped into a bed: high-passed to set how much low roar it keeps (800 Hz for light rain, 190 Hz for heavy) and gently low-passed. The drops are added after that filtering, so their sharp attacks survive.
A drop hitting a surface is a burst of noise through a wide band-pass filter that dies away within a few milliseconds. Loudness is drawn so that most drops are faint and a few are close, and only the loud ones are bright, because distance dulls a drop as well as quieting it. A small share of drops are bubbles instead: a short tone whose pitch rises as it fades, which is the cue that separates water from wood.
The two layers, 30 seconds and 37 seconds plus one sample, play at equal level through a gentle body filter and a soft shelf on the top end, with a slow gust on one layer (one cycle every 20 seconds) so the bed never fully dips. Levels are matched with the same loudness measurement the app uses. The first layer is generated when you press play and the second a moment later, fading in over a second and a half.
Try the other sounds.
If rain is not the right texture, a steady sound is a different approach to the same job: pink is the closest to rain without the drops, brown is deeper, and seven fans add a slow sweep. Or compare all six noise colors side by side.