The publication for AI music creators

The editorials are written by C. Niki Mclamb-Doanes — a perioperative registered nurse and the founder of Veridem Holdings Inc. — and they come from actually making the records, not from commenting on other people's. Recurring subjects: who owns a song when a model helped write it, what copyright currently recognises and what it doesn't, how taste survives a tool that will generate forever, and what the business of music is quietly doing about all of it.

Some essays run first on Substack; every one of them lives here in full.

Read the editorials ↗

This is the part of the site that exists nowhere else in this shape. A release page in The Drop carries the track, and underneath it the whole working file:

• The actual prompt used — the text that went into the model, printed as written, not summarised after the fact.
• The AI engine — which system generated the audio, and at what stage it entered the process.
• Tools used — every editor, plugin, studio and instrument that touched the track on the way to the master.
• How many generations were rejected — the count of takes that were made and thrown out before one was kept.
• What was changed by hand — the arrangement moves, re-sung lines, tempo and pitch edits and structural surgery done by a person after generation.
• The MHAIX authorship score — the track's rating on the MHAIX scale, broken out by dimension rather than given as a single number.
• What didn't work — the dead ends, in writing, including the versions that were better on paper than in the room.

The point of publishing all of it is simple: a song that shows its decisions can be argued with. One that doesn't can only be believed or dismissed.

Hear The Drop ↗

Organisers put up a source song along with its original prompt, lyrics and style notes. Entrants take it somewhere else and record how they got there: the new prompt, whether the musical style was revised, the structural changes, an AI critique, their own self-critique, and the final remix with its finished lyrics and prompt.

Completed entries are reviewed across five MHAIX authorship dimensions — Composition, Lyrics, Production, Curation and Iterative Labor — on a 0–40 scale. That review is an exposure to how MHAIX thinks about authorship, not an MHAIX score or certification. There is also a practice warm-up you can run end to end without submitting anything.

See the current challenge ↗

Separately, Spotlight runs a monthly video interview with an AI music artist, each with a plain breakdown of what the artist did and what the AI did.

It reads text — your lyrics, and whatever you write about the intent, the arrangement and the audience. It does not analyse uploaded audio files. That distinction is worth learning here rather than discovering it halfway through: no tool on this page listens to your mix. The read is about the writing and the thinking behind it, which is where most songs are actually won or lost.

It is free, limited to two reads per visitor per day, and nothing you paste is stored. If you want a lyric workspace that keeps your drafts and your decisions, that's Re:Writ.

Open the Listening Room ↗

Type an artist, a mood, a time of day, a weather system, a city. Set the genre dial, the Mild-to-Chaos variation slider and the tempo, or leave them on Auto and let the genre choose. Spin the record and a brief comes out: texture, tempo feel, drums, a scene, arrangement instructions, then the genre and BPM.

It runs entirely in your browser, costs nothing and there is no limit on how many times you can spin it. To turn one of those briefs into a saved blueprint with documented decisions, that's Stem:Ample.

Spin 8-Ball Vinyl ↗

You build a track across a timeline board from genre, instrument and wild cards while a B-Side opponent races to claim the same song. There's a timer, a bank, effects and Ghost Tracks, a daily challenge and a leaderboard. The result maps to the MHAIX authorship scale, so the game is quietly making the same argument the rest of the site makes: what you decided is the thing being scored. Free to play in the browser.

Play Overdub ↗

Written breakdowns can only carry so much. The video page carries the rest — a take being rebuilt, a decision being made out loud, the version before the version that got released.

Watch In the Room ↗

They are disclosed as personas everywhere they appear. A release credited to one of them still carries the same documented production breakdown as everything else in The Drop, and the same AI disclosure applies. For releases assembled as an EP, the sequencing and release decisions are documented in Compose:EP, live now at composeep.com.

Virtual artist · 01

Dizzy Rey

Dizzy Rey is one of the virtual artists behind Rendered Sound's releases — a persona with her own voice, taste and point of view, steered by a human at the console. When a track carries her name, the decisions behind it are documented the same way everything else here is: out in the open.

Hear her on the Music page

It covers the music, the artwork, the writing and the tools, and it is written to be read by a person rather than a lawyer. If you want to know what a human did on any given piece of work here, the disclosure is the starting point and the per-release breakdowns are the detail.

Read the AI disclosure ↗