Jelle Akkerman is an engineer and artist based in Northern Italy. In June 2026 he launched the first two plugins in his every something series: Artikulation and Glosso. Glosso is the instrument — a vocal generator that requires no input. You load it into a DAW track, press play, and it begins to speak. Two voices surface from vowel filters, drifting through formant space, following rhythms that gather and dissolve. It sounds like a voice, but not like any particular voice saying anything particular. It has the form of language without the content of language.

Glosso's engine is the rungler circuit from Rob Hordijk's Benjolin synthesizer. The rungler couples two oscillators through a shift register in a feedback loop: one oscillator clocks the register, the other's output is sampled and fed back as control voltage into the first. The resulting behavior is neither random noise (it has internal structure) nor stable deterministic oscillation (it evades any single directing intention) — it is pseudorandom: organized but unpredictable, logical but not commandable. Hordijk called the Benjolin a "self-playing instrument": you don't play it, you set its conditions. Glosso transplants this circuit logic into vocal synthesis, running two rungler-style chaotic circuits through formant filters and letting them roam through vowel space.

The entire history of vocal synthesis is a history of control. From the Klatt formant synthesizer of the 1980s through Vocaloid through current neural text-to-speech, the whole lineage is driven by one impulse: make the machine say what you want it to say with increasing precision. This history has built the already-construct: voice synthesis equals directed utterance; a synthetic voice is a tool and proxy for human intention. Whether it's a movie score or a voice assistant, this logic has never been questioned, because "make the machine speak for you" is the entire premise.

Glosso chisels through this. It does not build a mouth that can be directed. Instead, it removes the possibility of semantic control from the design itself: two circuits feed into each other, and no single external intention can intervene in what they say. What is chiseled away is the command over what gets said; what remains is the phonemic form itself — vowel contours, resonant-cavity rhythms, consonantal pulses — roaming freely without a speaking intention to drive them. The name points directly at this structure: glossolalia, the religious phenomenon of producing sounds that have full phonemic form but constitute no known language. Linguistically, glossolalia is the remainder of language-form after semantic content has been removed — the skeleton of utterance without the flesh of meaning.

To understand why now, consider the rungler logic's own situation. Hordijk died in 2019, and the Benjolin is entering historicization: commercial clones, museum acquisitions, masterclasses, academic papers, Eurorack "tributes." The rungler is in the process of shifting from "a still-growing structural logic for organizing analog chaos" to "Hordijk's technique" — named, attributed, crystallized as a method under a specific person's name. Once that naming fully hardens, the rungler moves from living remainder to sediment of the already-construct. Glosso, before that hardening completes, carries the rungler logic somewhere it has never been — vocal formant space — and makes the logic live again on unfamiliar ground, producing something it has never produced before.

Glosso's naming gap reveals its remainder position. It is not a vocal synthesizer (it cannot be directed to any phoneme), not a noise generator (it has phonemic structure and rhythm), not a formant filter (it generates rather than processes), not a granular synthesizer (its temporal structure is circuit-based, not grain-based), and not a random signal source (it has internal consistency). The every something site describes it in one line: "your computer speaks in tongues" — the name is the experience, not a category. Two more plugins are due in July 2026: perp (a Risset rhythm live looper) and otophon (a psychoacoustic tone generator). The logic is still growing. Jelle Akkerman has no art-world position.

Han Qin (秦汉) · @hqinjarsy · SAE Judgment & Aesthetics