A second complete media space

Every Sound

Pixel Theory is not about images. Images were just the first instance. Digitization makes every medium a finite, fully addressed set — and here is the second one: sound.

Ten seconds of CD-quality stereo audio is 882,000 samples of 16 bits each. The set of every possible such clip therefore has 2^14,112,000 members:

…

= 2 ^ 14,112,000 — every possible 10-second clip at 44.1 kHz, 16-bit stereo

This set contains every sentence ever spoken by every human, in every voice, in every language — including all the ones never uttered. Every song, every performance of every song, every mix that was never released. Your own voice saying words you will never say. All of it fixed, finite, addressed — and, like the image space, overwhelmingly made of hiss.

Tune in

Uniform random samples from the set. Almost every clip you hear has never been heard by any ear in history, and after it plays, it never will be again.

You are the only ear that will ever hear this. When it ends, its address returns to the dark with the rest of the 10^4,248,135.

Address a recording

Drop in any audio file — your voice, a song, a memo. It decodes to PCM samples, and those samples are a number: the address the recording has held since before sound existed. A microphone, like a camera, looks things up.

drop audio here, or click to choose · processed locally, never uploaded

The pattern generalizes

Text was first — Borges wrote the library, and the internet built it. Images are the Javen Number. Sound is this page. Video is a path through the image space, one frame per step. In every case the same facts hold: the set is complete and timeless; creation is navigation; meaning is compressibility, provably of measure zero in its ocean; and truth about origin lives not in the samples but in the witnessed event that selected them. Whatever reality is, it is not the set of its possible appearances — it is the process that chooses among them.