No. 0398
A pianist exhales twice between phrases; the second exhale carries more dust.
III · MMXXVIa quiet research notebook for an experimental machine-listening lab housed in a converted analog telephony exchange
Listening occurs upstream of meaning.
We work upstream of language. Before a phrase becomes a meaning, before a meaning becomes a stance, there is a long stretch of unsorted air — pressure, harmonic, breath. Our instruments rest in that stretch, attentive to the textures that ordinary listening discards.
The lab occupies the third floor of a 1932 telephony exchange. The relays were stripped out in 1978; the racks remain, repurposed now as scaffolding for our recorders. The walls retain the carbon dust of half a million switched calls.
Our principal subject is weather in voice — the slow shifts in fundamental frequency, breath rate, and timbral wear that mark a speaker across a single afternoon. We do not transcribe what is said. We chart the climate of saying it.
Each listening session is preserved on quarter-inch magnetic tape and on a parallel digital recorder. The tape is the record of record; the digital is the working surface. We trust neither alone.
Three days a week we are open to visiting researchers. The other four, the rooms are silent except for the slow hum of the dehumidifier and the occasional click of a relay we left in place for company.
Sessions run between forty minutes and four hours. Nothing is paused; nothing is retaken. The tape rolls until the question rolls out of it.
Quarter-inch tape and a 24-bit digital recorder run simultaneously, clocked from the same crystal. Discrepancies between the two are themselves the subject of a side study.
We do not produce text from speech. We produce charts of breath, intervals, and timbral wear. The words are stored, sealed, and returned to the speaker if requested.
Each recording is followed by an equal length of silence in the same room before the next session begins. The room must reset.
The instrument hears the room. The room hears the instrument. We listen to the conversation between them, and write very little down.
The building was constructed in 1932 as the No. 4-Ki telephone exchange and decommissioned in 1978. The carbon-rod relays were carted out in nineteen separate truck loads. We arrived in 2019 and have been listening since.
We are a small team — four full-time researchers, two visiting fellows, and the building, which we count as a colleague. Most of our findings are slow, undramatic, and not yet ready to be reported. We are in no hurry.
Correspondence is welcomed by post. Reply is unhurried.