The work

You will listen to AI-generated audio and judge it the way a producer judges a rough mix — closely, repeatedly, and with reasons. Most tasks are head-to-head comparisons: two outputs from the same prompt, and you decide which is better on musicality, creativity, prompt adherence, vocal quality, and production/mix quality, then write a short justification in audio-engineering terms. A second stream of work is structured annotation: tagging genre and sub-genre, marking song structure (intro, verse, pre-chorus, chorus, bridge, outro), describing vocal timbre and instrumentation against a fixed taxonomy, and flagging artifacts like pitch smearing, phase issues, clipping, or unnatural breath placement.

The German requirement is substantive, not cosmetic. You will assess whether generated lyrics scan naturally in German — stress patterns against the beat, rhyme quality, idiom, register, and whether a sung line survives compound words and umlauts without turning to mush. Prompts and task instructions arrive in English, so you need to work comfortably in both languages.

What Mercor screens for

  • Verifiable production, engineering, or mixing history — credits, releases, studio work, or a portfolio you can point to
  • Formal or demonstrable grounding in theory and performance, often a music degree
  • Evidence you have written or produced lyrical music in German, not just listened to it
  • Consistency: whether your ratings hold up across repeated and near-identical items, and whether your written rationales match the scores you gave
  • Vocabulary precision — reviewers notice the difference between "sounds bad" and "the vocal bus is over-compressed and the sibilance is untamed"

Logistics

Fully remote and asynchronous. Commitment is part-time at 10+ hours per week, with an immediate start and a project duration of up to six months. Onboarding runs through a resume upload, an application form, and a 25-minute conversational AI interview covering your background and motivations; next steps typically follow within a few days. Observed pay for this listing is $32.48–58.46/hr — the band reflects what the platform has posted, not a guarantee, and placement within it depends on assessed depth and task type. Good headphones or monitors and a quiet listening environment are effectively required.