What the work involves
You work across two audit collections. In the Multilingual Transcription Audit, you listen to a recording of human or agent speech in Spanish, read the transcription an annotator submitted, and decide whether it faithfully captures what was said. You tag every applicable error code — not only the first one you notice — assign pass or fail, and write a short rationale in English that leads with the critical error so a reader understands the verdict immediately. In the Forced Alignment Audit, you review audio already segmented at the word level by a machine and verify that each start and end timestamp lands where the word actually begins and ends, and that the reference form matches what was spoken. Where alignment is wrong you adjust boundaries, split, merge, insert, or delete segments.
Both collections are budgeted at 5.00 audit hours per task, which tells you what kind of work this is: careful, evidence-backed listening rather than throughput. You are not producing transcripts — you are judging someone else's, and the written reason matters as much as the verdict.
What the platform screens for
- Variety-specific fluency. Native or near-native Spanish as spoken in Spain — born and raised in a region where that variety is dominant, or fully fluent with five or more years of residence there. This is a hard gate, and screeners probe it with questions about regional phonology and informal speech, not just self-report.
- English rulebook literacy. Conventions, error codes, and all rationale are in English. Expect to demonstrate that you can read a dense written standard and write a compact, unambiguous justification in it.
- Rubric discipline over instinct. The decisive signal is whether you apply a written rule the same way on judgment 300 as on judgment 3, and whether you can separate genuine acoustic ambiguity from a clear annotator mistake.
- Prior transcription, subtitling, localization, linguistic annotation, or QA work is a strong signal. Phonetics or linguistics training and familiarity with waveform tools help but are not required.
Logistics
Remote and asynchronous, with a short calibration phase before any production work is released. Pay has been observed at $40/hr for this collection; rates on Mercor are set per project and are not guaranteed. Volume depends on collection demand, so treat it as flexible contract work rather than a fixed schedule — you will need reliable headphones, a quiet listening environment, and enough uninterrupted blocks to sustain attention across long audio.