Why This Archive Does Not Use MP3 Files

The site describes 10,000 soundscapes, and none of them exists as an audio file. Every track is generated in the browser when you press play. That decision shapes everything else about the archive, from page weight to how a seed can be documented.

The storage problem

Ten thousand tracks at a conservative five minutes each is roughly 35 days of audio. At a decent bitrate that is hundreds of gigabytes, before considering that each track would need multiple formats and bitrates for different connections. Serving that library is a full-time infrastructure project, not a website.

Generating the audio removes the storage problem entirely. The page contains the instructions for making the sound, not the sound itself.

Why the same seed always sounds the same

A generated track could easily be random every time, which would make the archive meaningless. Here the seed number drives a deterministic random generator: the same number produces the same parameter set, which produces the same audio. That is why a page can state a tempo, a filter value and a tuning and have those numbers remain true.

What you give up

Procedural audio is not a recording. It cannot capture the specific room, the specific instrument, or the performance of a musician. A felt piano here is a model of a felt piano, not a sample library. If you need the sound of a particular grand piano in a particular hall, this is the wrong tool.

What you get instead is consistency and instant start. There is no buffering, no download prompt, and no difference between the first play and the hundredth.

Copyright and licensing

Because nothing is sampled from a third-party recording, there is no master recording to license. The audio is produced by the code on the page. That simplifies use for personal listening, though it does not automatically grant rights to redistribute the site’s code or branding.

What this means for you

You can open a seed, press play, and hear the same thing on a phone, a laptop or a desktop, as long as the browser supports the Web Audio API. You can also read exactly which parameters produced it, which is not something a file-based library usually offers.

Open the player and try a seed, then read its parameters on the seed page.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top