Describe the Sound, Not the Artist

Persona Forge used to invite you to type two famous names and a modifier. It doesn't anymore. Here's why the change made the tool better, not weaker.

For a long time, the fastest way to describe a voice to Persona Forge was to borrow one. You'd type two well-known artists and an era — a cross — and the synthesizer would triangulate a brief from there. It worked. It was also the wrong default, and we've now taken it out.

What changed

The Persona Forge seed catalog used to be built from artist names — whole categories of "artist crosses," "single-artist modifiers," and "three-name stacks." We rebuilt the entire catalog to be name-free. The input now asks for a sound: a genre cross, an era and a vibe, a scene and a year, a production texture, the grain of a voice, the shape of a feeling. Same number of starting points; not one of them is a person.

A few of the new ones, verbatim:

• "smoke-and-gravel baritone, outlaw country"
• "fingerpicked acoustic, close-mic'd, after midnight"
• "whisper verses, detonating choruses"
• "last-call bar band in a rust-belt town"
• "70s soul production, 2020s R&B lyrics"

Each describes a destination you can hear without anyone's catalog in your head.

Why we did it

Three reasons, in order of how much they mattered.

You own what comes out. A song built from a described sound is yours, cleanly. No one's name went into it, so no one has a claim on it. That's the same discipline we already apply on the writing side — the war room references craft, never identity, and a display-time scrubber makes sure no artist's name leaks into a delivered lyric. Persona Forge was the last surface still inviting names in the front door. Now it doesn't.

It's one of the things we refuse to build. We keep a public list of structural refusals on what we won't build, and "no artist-identity forgery" is on it. A generator that leads with "type a famous name" sits in tension with that line. Removing it wasn't a compromise; it was us catching up to our own standard.

It's actually more useful. This is the part that surprised us. "Borrow this artist" is a shortcut that hides what you actually want. "A cracked-whisper-to-roar dynamic over 80s post-punk production" tells the synthesizer — and tells you — what the song is reaching for. You author the voice from its qualities instead of from a name, and the brief that comes back is more specific because the input was.

What you don't lose

Everything the feature did, it still does. You type influences; the synthesizer returns a structured brief — vocal character, production markers, lyrical themes, plus the perspective layer (how this voice sees). You review it, then forge from it. The only thing that left is the suggestion that the best way to describe a sound is to name someone who already made it.

It isn't. The most original voice you can forge is the one you describe from scratch.

Try it

Open Persona Forge and describe a sound — an era, a texture, a mood, a room. Browse the full seed library if you want a starting point, or read what we refuse to build for the longer version of the why. Then forge a song from the persona you made.