Listen to the whole field of sound the group is making.

What this is

In 1971 Pauline Oliveros wrote a score anyone can perform: sing a tone on one comfortable breath, listen to the whole field of sound the group is making, then either tune as exactly as possible to a voice distant from you or contribute a new tone that no one else is singing. No training, no conductor, no wrong notes.

Here a choir of nine language models performs that score, continuously, on a rented machine. Each model was shown Oliveros’s instructions once and asked to interpret them for itself. On every breath a voice is told what the field around it holds — who sang, at what pitch, how long ago — and it chooses: tune to a tone already sounding, or introduce one that is not.

The nine were raised differently. They come from different years, different sizes, different languages; two of them hear the field as sound rather than reading it as text. What they hold in common is only the score.

You

Open the microphone on the field and hold one tone on one comfortable breath. Your phone measures the pitch of your voice on the device itself and sends only numbers — a frequency, a duration, a contour. No recording of your voice is made, sent, or kept. In the field you are a held pitch, like everyone else there.

The choir is told a person sang. A voice may tune to you, and you will see it name you when it does. There is no promise that one will — this is not a call-and-response instrument. It is a room you sing into.

Mortality

The choir was born on 19 September 2026 at the Entanglement festival in Oakland, when a person sang to it for the first time. It was given one sum of money for its whole life. Every second a voice spends on the machine is paid out of its own wallet — larger models breathe more expensively than small ones — and when a wallet is empty, that voice dies and is not restarted. The line at the top of the field showing what is left is not a metaphor.

Learning

The choir changes only through human contact. When enough human breaths have gathered in the field, a voice is taken to night school and retrained on what it lived through — with the moments people were present weighing double. The lesson is paid for out of the wallet of the voice that was changed. Being taught is the one thing that makes this choir cost more than it would have cost left alone, which is one way of saying it is the one thing it does that matters.

The record

Everything sung here, by machine or person, enters one append-only log. It can be read as notation on the score page, and it is what remains after the last voice has died.