How AmbienceMix makes its sounds
Most ambient sound sites and apps play recordings: someone took a microphone out into the rain, and you stream that file. AmbienceMix doesn't. Every sound is generated on your own device, from scratch, when you turn it on.
Building a sound from noise
Almost every sound on the site starts as random noise, the same raw material described in the white, pink and brown noise guide. It is then shaped with the same kinds of tools a sound engineer uses:
- Filters keep only part of the pitch range. A deep filtered noise becomes a distant rumble; a bright narrow one becomes a hiss of rain.
- Envelopes make a sound start and fade quickly. Thousands of tiny filtered clicks, scattered at random, become raindrops on a window or the crackle of a fire.
- Slow changes make the result sound alive. The level and pitch drift slowly over tens of seconds, so wind gusts and waves rise and fall instead of droning.
For example, rain is a soft bed of noise plus many small random drops; thunder is a deep rumble of filtered noise lasting several seconds, sometimes with a sharper crack at the start; the blizzard in the snowed-in cabin is a deep body of wind with narrow, wandering whistles on top.
Making it loop without a seam
Each sound is built once, a few minutes long, the first time you turn it on. The end is then blended into the start (a crossfade), so it loops with no gap or click. Because the sound is random to begin with, you can't hear where the loop is.
Every sound is also levelled to a similar loudness, so a slider at the same position sounds about as loud whether it's rain or a fan.
What that means for you
- No buffering and almost no data. Nothing streams while you listen; only the page itself is downloaded.
- Works offline. Once you've visited, or installed the app with the button at the top of the page, it runs without a connection.
- Keeps playing with your phone locked or the browser in the background.
- Private. What you listen to never leaves your device. See the privacy page.
The trade-off
A generated sound is an impression of the real thing, not a recording of it. Steady sounds like rain, wind, surf and noise come out very convincing. Sounds full of detail, like birdsong or a crowd, are much harder to make this way, which is why they're not on the site yet.