About

They Speak is a listening post for non-human communication. Each entry presents a recently-decoded finding from an animal species not as a paper to read but as something you can hear, sort through, and try to speak back to.

The site is built on public scientific research. The audio is synthesized from real timing or spectral descriptions, not recordings — a sonification of structure. The text never claims more than the science does.

New species are added when the underlying research is solid enough to demonstrate, not when it is interesting enough to claim. The list will grow slowly.

How this site works

  1. Source. Every entry starts from a specific peer-reviewed paper. Where the authors released raw data, we mirror it and join across datasets. Where they didn’t, we author representative samples from the published descriptions — and say so in plain language on the page.
  2. Synthesize. Each utterance is reconstructed with Web Audio from timing or spectral descriptions. Rhythm and sequence are real. Timbre is approximated. It is a sonification of structure, not a recording.
  3. Display. Plain SVG, plain CSS, no recordings. What is on screen — a waveform, a dance, a classifier card — is always derived from the dataset it names.
  4. Listen. You hear what they heard. You match what they said. We do not translate.

Working principles

  1. Do not claim to translate.
  2. Do not anthropomorphize.
  3. Cite the actual paper.
  4. The gap between hearing and understanding is the point.