Sonic Audit Specialist - Italian at Mercor

Sonic Audit Specialist (Italian) | $39.50/hr | Worldwide Remote

Mercor is seeking native and near-native Italian speakers to audit AI training data for Amazon's Sonic collections. This is careful, detail-oriented listening and judgment work — not content production. You will evaluate transcription accuracy and forced alignment quality, applying structured rubrics and writing clear English rationale for every decision.

About the Role

You will work across two Amazon Sonic audit collections:

• Multilingual Transcription Audit: Listen to recordings of human or agent speech, review annotator-produced transcriptions, and judge whether they accurately capture what was said. Apply a fixed set of error codes, assign a pass/fail verdict, and write a concise rationale for each decision.
• Forced Alignment Audit: Review audio segmented at the word level by an automated system. Verify that start and end timestamps correctly match each spoken word, and that reference transcriptions match the spoken form. Correct boundary errors by splitting, merging, adding, or removing segments as needed.

Both collections run at 5.00 audit hours per task. Every judgment requires a written explanation, making this precision-focused work rather than high-volume output.

What We're Looking For

• Native or near-native Italian fluency as spoken in Italy — either raised in a dominant Italian-speaking region or fully fluent with 5+ years of residence there. This is a hard requirement.
• Strong English reading and writing skills — all conventions, error codes, and audit rationale are written in English.
• Disciplined rubric application — ability to apply written standards consistently across hundreds of judgments. Prior experience in transcription, subtitling, localization, linguistic annotation, or QA is a strong advantage.
• Careful, attentive listening — able to distinguish similar sounds, identify word boundaries in continuous speech, and differentiate acoustic ambiguity from clear annotator error.

Nice to Have

• Prior AI data annotation or model evaluation experience
• Background in phonetics or linguistics
• Experience with audio editing or waveform tools

How It Works

You listen, judge, and submit. When errors are found, you tag all applicable error codes (not just the first), and explain in plain language why a task passed or failed — leading with critical errors so the reason for failure is immediately clear. A short calibration phase is required before production work begins.