What the work actually involves
This is timbre identification at volume. You listen to short clips — often isolated notes, phrases, or brief passages — and tag two things: the instrument family (strings, brass, woodwind, percussion, keyboard) and the specific instrument within it. Turing supplies reference audio and an annotation guide; your job is to apply them consistently rather than creatively. Expect edge cases that are the whole point of the project: viola versus violin in a middle register, cor anglais versus oboe, a synth patch imitating a brass section, a processed or reverb-heavy recording where the attack transient is the only reliable cue.
The project window is two days with flexible hours inside it, and the posted band of roughly $150–$160 is what a full push across both days has been observed to pay — it is a project total, not an hourly rate, and the platform has not confirmed the period. Strong performers may be invited to continue on further audio work.
What the screen is looking for
- A verifiable music background. A bachelor's in music or a closely related field, plus three or more years in performance, production, audio engineering, or comparable work.
- Ear accuracy under time pressure. The ~15-minute qualifying assessment is a listening test. It decides onboarding.
- Guideline discipline. Annotation projects are graded on agreement with a rubric and with other annotators, not on musical sophistication. Someone who overrides the label set because they hear something more interesting is a liability.
- Tolerance for repetition. Hundreds of short clips in a sitting is the job description, not a hazard of it.
Logistics
Fully remote and asynchronous within the two-day window. You need reliable headphones or monitors and a quiet environment — laptop speakers will cost you on quiet or low-register clips. Payment is in USD through Turing's contractor pipeline. Bonus credit for experience with transcription, remixing, music information retrieval, or prior AI/data annotation work.