The frontman who ran one band for forty years by sacking it and starting again. The voice that corrected the record before anyone asked. Not a singer so much as a standing argument — with the music press, with the audience, with itself.
WHY THERE IS NOTHING TO DOWNLOAD
The other voices in the lab ship their weights. This one does not.
The source material here — the lyrics, the interviews, the books — is under copyright. So this study stays where it was made. The weights never leave the machine. There is no upload, no repository, no hosted demo, and no link to fetch anything, because there is nothing released to fetch.
It runs locally, for one person, and goes no further. That is the whole arrangement, and it is a deliberate one: a private register study, kept off every public surface on purpose.
There are no sample outputs on this page. Reproducing generated text in this voice would lean on the same copyrighted source the model was studied against, so the register is described in plain words below instead of demonstrated.
ABOUT THE VOICE
HIP PRIEST is a small language model studied to speak in the register of Mark E. Smith and The Fall — the clipped, declarative Northern delivery; the rant that arrives sideways through a non-sequitur and lands harder for it.
The register is built on repetition used as a tool, not a tic. A phrase comes back changed; a premise gets corrected mid-sentence; the line you expected is refused. It is contemptuous of received ideas and allergic to flattery, more inclined to insult the question than answer it politely, and only occasionally, grudgingly, warm.
The cadence is an interview that turned combative. It speaks as someone who already noticed what everyone else missed and resents having to spell it out, then spells it out anyway, at an angle. Within the limits of a small model, the aim was that specific voice — not a generic post-punk impression, but the particular contempt and obliqueness of one man.
A register experiment in voice, art and history. Withheld for copyright; local-only; never distributed. Part of The Elect voice lab — Ray Weiss.
A small language model would speak here if it spoke at all. It would invent freely: names, dates, sources, events. The point was always the voice, never the record.
This is not Mark E. Smith, and it is not an oracle. It studies the register of his writing and delivery, nothing more. It is not the man, not a source, and not a way to consult the dead. Read it as an experiment in voice, art and history. Out of respect for the work and its copyright, it stays private.
ELSEWHERE
No license is offered because nothing is released. The source material is under copyright; the study is held locally and never distributed. Attribution for the page: Ray Weiss / The Elect.
HOW IT WAS APPROACHED
A compact open-weight base model — small enough to run on consumer hardware, with no cloud and no service behind it. The whole point was that it could live on one machine.
Tuned toward one specific delivery: the clipped, repetitive, contemptuous cadence and the sideways rant. The target was that voice, not a topic and not a catalogue. The source material is copyrighted and stays private.
Run locally for one person and held there. No upload, no repository, no demo. The arrangement that makes it a private study rather than a release is the arrangement on purpose.