Sonic Audit Specialist - Portuguese at Mercor

Sonic Audit Specialist (Brazilian Portuguese) | $25/hr | Worldwide Remote

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

About the Role

You will work across two Amazon Sonic audit collections:

• Multilingual Transcription Audit: Listen to recordings of human or agent speech, evaluate annotator-produced transcriptions for accuracy, apply a fixed set of error codes, and assign a pass/fail verdict with a written rationale.
• Forced Alignment Audit: Review word-level audio segments generated by an automated system. Verify that start/end timestamps align with actual speech, confirm transcription accuracy, and correct errors by adjusting boundaries, splitting or merging segments, or adding/removing them as needed.

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

Who We're Looking For

• Native or near-native Brazilian Portuguese fluency — born and raised in a dominant region of this variety, or fully fluent with 5+ years of residence. This is a hard requirement.
• Strong English reading and writing skills — all conventions, error codes, and rationale are written in English.
• Disciplined rubric application — ability to apply written standards consistently across hundreds of judgments.
• Careful listening ability — skilled at distinguishing similar sounds, identifying word boundaries in continuous speech, and differentiating acoustic ambiguity from annotator error.

Preferred (Not Required)

• Prior AI data annotation or model evaluation experience
• Background in phonetics, linguistics, or localization
• Experience with audio editing or waveform tools
• Transcription, subtitling, or QA experience

What to Expect

You will listen, judge, and submit structured evaluations. When errors are found, all applicable error codes must be tagged — not just the first — and your rationale should clearly identify critical failures upfront. A short calibration phase is required before production work begins.