Brainy Records

Built for Show HN

A record-label simulator for fictional AI bands.

Brainy Records takes a genre prompt and generates the band around the song: name, members, lore, album cover, and a vocal track. The goal is not just another song generator; it is a shareable fictional act with music attached.

Listen before trying it

These are real published bands from Brainy Records. Each one includes a generated concept, cover, and playable song.

VANDAL AMOUR cover
VANDAL AMOUR

A hybrid of contemporary metalcore and alternative pop, featuring the cinematic production of Sleep Token, the…

Chrome Sonata cover
Chrome Sonata

A cinematic blend of Neoclassical structures and Cyber-Techno. It features virtuosic string solos layered over…

Satellite Sunsets cover
Satellite Sunsets

A high-octane blend of contemporary indie-folk and bubblegum-pop song structures. The sound features driving a…

What it generates

Band identity

Name, tagline, genre, member roster, personalities, and backstory generated by Gemini.

Visuals

Album covers and member portraits generated with Imagen, stored and reused on server-rendered band pages.

Vocal music

Lyria 3 generates the paid full vocal song path. Realtime/instrumental previews stay on the cheaper model branch.

Technical notes

Model branching

The backend branches by Lyria model ID. Lyria 3 uses REST generateContent with inline MP3; realtime previews use the websocket Live API path.

Credit accounting

Paid AI calls reserve a credit in Firestore before the model request. Infra failures refund; safety-filter rejections do not, because the upstream call still costs money.

SSR for crawlers

Band pages, genre pages, pricing, blog, and this page are server-rendered from Express. The SPA owns creation and studio workflows.

Honest state

  • 14 published bands, mostly seeded.
  • One free vocal song per Google account.
  • $0.99 for one more full song; packs start at $4.99.
  • Zero completed paid orders so far; this launch is a funnel test, not a victory lap.
  • Lyria 3 vocals are useful for ideation and novelty, not radio-master quality.

Useful feedback

The most useful feedback is where you lose interest: before trying it, after hearing samples, at sign-in, at pricing, or at PayPal. That tells me whether the issue is positioning, product value, trust, or checkout friction.