Skip to content
Back to corpus
Paid rater program

$500 per quarter to rate AI lyrics against a real rubric.

We're onboarding 10 working songwriters as paid rubric calibrators. Your job: rate ~100 AI-generated lyric outputs per quarter against the published Lyric Scoring Standard. Async, remote, paid quarterly.

$500

per quarter (~$5/rating)

~100

lyric ratings per quarter

Async

3-5 hrs/quarter, when you want

Who we're looking for

  • Working songwriter with at least one cut on a major-label release in the last 5 years (BMI/ASCAP/SESAC repertory verifiable)
  • Comfortable rating against a structured rubric (we provide the 12-metric scoresheet; you score 0-100 per metric with a 1-line rationale)
  • Comfortable signing a CC BY 4.0 contributor agreement so your ratings ship as part of the public corpus (your name optional; default is anonymized)

What you actually do

  1. Receive a private rating queue at the start of each quarter (~100 lyrics).
  2. For each lyric: 0-100 score per metric (12 metrics) + a 1-2 sentence rationale on the strongest and weakest metric.
  3. Optional: flag up to 5 lyrics per quarter for the “transcendent line” corpus — the line you wish you'd written.
  4. Quarter close: $500 paid via invoice or Stripe. Your aggregate ratings + flagged lines feed the next rubric calibration cycle.

The honest pitch

The rubric is published, but it's been stress-tested mostly by AI evaluators and one founder who wrote songs in college. Without working-songwriter ratings the corpus calibrates against itself — circular validation. You're the external pressure. The pay reflects that. Long-term, this corpus becomes the largest expert-rated AI-lyric calibration dataset in the world; your name (if you opt in) gets cited every time it's referenced.

Apply

Email raters@songforgeai.com with:

  • Your name + a link to one cut you're proudest of (Spotify, Apple Music, or registry link)
  • One paragraph: why the rubric will be better with your ratings in it
  • Your preferred quarterly pace (~100 ratings is default; we're flexible)

We respond within 5 business days. If it's a fit, the first batch lands within 7 days of the contributor agreement. The first 10 raters get founding-rater status (cited in v1.2.0 of the rubric publication).

Apply via email