Sonic Audit Specialist - Korean (Remote)
Listing checked September 18, 2026 · pay as published by Mercor
Overview
Mercor seeks native and near-native Korean speakers to audit AI training data rather than create it. You will work on two Amazon Sonic collections: Multilingual Transcription Audit and Forced Alignment Audit. In the first, you compare a recording of human or agent speech with an annotator's transcription, apply set error codes, and issue a pass or fail verdict with a brief rationale. In the second, you check word-level segments from an automated system, verify start and end timestamps plus reference text, and fix boundaries by splitting, merging, adding, or removing segments. Each task runs at 5.00 audit hours and every judgment needs a written reason, so the role favors careful listening over speed.
What You'll Do8
- 1Listen to recordings of human or agent speech and compare them with annotator transcriptions.
- 2Apply the fixed set of error codes, assign a pass or fail verdict, and write a short rationale for each decision.
- 3Review word-level audio segments that an automated system has produced in forced alignment tasks.
- 4Verify that start and end timestamps match where each word begins and ends, and that reference text matches the spoken form.
- 5Correct machine mistakes by adjusting a boundary, splitting a segment, merging two segments, adding a missing segment, or removing a segment with no word.
- 6Tag every applicable error code, not just the first one you notice.
- 7Explain in plain language why a task passed or failed, calling out critical errors first.
- 8Complete a short calibration phase before production work starts.
Requirements7
- 1Native or near-native command of Korean as spoken in South Korea. You either grew up in a region where that language and variety dominate, or you have five or more years of residence there. This is a hard requirement.
- 2Strong English reading and writing. All conventions, error codes, and audit rationale use English, so you need to read a detailed rulebook and write clear feedback in it.
- 3Discipline applying a written rubric across hundreds of judgments instead of relying on instinct.
- 4Prior work in transcription, subtitling, localization, linguistic annotation, or QA is a strong signal.
- 5Careful listening. You can tell similar sounds apart, find where one word ends and the next begins in continuous speech, and separate genuine acoustic ambiguity from a clear annotator error.
- 6Helpful but not required: prior AI data annotation or model evaluation work, a phonetics or linguistics background, and experience with audio editing or waveform tools.
- 7Ability to tag every relevant error code and explain pass or fail decisions in English.
Who Should Apply
The ideal candidate speaks Korean as used in South Korea at a native or near-native level and can write clear English feedback from a detailed rulebook. You have audited or annotated speech before, whether in transcription, subtitling, localization, or forced alignment, and you can apply the same rubric across hundreds of tasks without drifting. This role suits someone who treats each 5.00 audit hour as careful listening work and documents every decision. Candidates who lack the South Korea language variety may find the role less suitable, as may those who prefer fast volume annotation over written rationale. Common reasons for rejection include failing the Korean residency or fluency bar and submitting rationales that miss the written rubric, such as tagging only one error when several apply.
Salary Insight
Pay appears as $25.00 per hour for remote audit work. The rate usually matches candidates who bring some transcription, localization, or annotation QA experience, and the listing does not tie it to a named seniority band. Mercor pays by the hour, with calibration before production, so the figure covers completed audit hours.
Pay and demand for Language & Linguistics roles
AggregatedTypical pay
$30/hour
This role
$25/hr
Most Language & Linguistics roles pay $22–$48 per hour. This role's pay falls inside that range.
Based on 257 similar roles that publish pay · 101 publish only a top rate; those count at the rate they gave
Rates shown per hour. Yearly and monthly pay converted; one-time fees and non-USD pay are not included.
- Live similar roles
- 295
- Listed in last 30 days
- 196
- Remote
- 100%
Hiring most right now: SME Careers (105) · micro1 (77) · Mercor (37)
Most requested skills · share of roles
- localization32%
- translation quality review29%
- translation27%
- proofreading27%
Figures from Language & Linguistics roles live on NearSkill when this page loaded. A role can close before you apply, so check the listing itself.
Compare your resume against these rolesLocation
Compensation
$25/hr
Required Skills
Application Tip
State your South Korea connection in the first line of your application, including where you lived and for how long. Add a short writing sample in English that tags multiple error codes in a transcription and gives a clear pass or fail rationale. That shows the two hard filters: native Korean as spoken in South Korea and disciplined rubric application.
See NearSkill jobs more often in your search
Application & verification flow
1Instant rubric match
Your resume is scanned against this role’s requirements to check qualification fit.
2Screened before the employer sees it
Only profiles that clear screening are passed on.
3Outcome by email
We notify you at the address on your resume once the screening is reviewed.
Similar open positions
Explore active roles that match your skills and interests.
Mercor
VerifiedMandarin Chinese Sonic Audit Specialist (Remote)
Mercor seeks native and near-native Mandarin Chinese speakers for a remote audit role, not a content production job. You join two Amazon Sonic collections: a multilingual transcription audit and a forced alignment audit. In the first, you compare an annotator's transcription against recorded speech, apply error codes, and give a pass or fail verdict with a written reason. In the second, you inspect word-level audio segments, verify start and end timestamps, and repair boundaries when the system misplaces a word. Each task runs at 5.00 audit hours, and a short calibration phase comes before production work.
Mercor
VerifiedJapanese Sonic Audit Specialist - Remote
Auditors in this Mercor role check Japanese AI training data instead of creating it. You work inside two Amazon Sonic audit collections: the Multilingual Transcription Audit and the Forced Alignment Audit. In the first, you compare recorded speech with an annotator's transcription, assign error codes, and write a brief pass or fail rationale. In the second, you inspect word-level audio segments, confirm start and end timestamps, and fix boundaries when the automated split misses the spoken word. Each task runs at 5.00 audit hours, and a calibration phase comes before production work.
Mercor
VerifiedSpanish Sonic Audit Specialist - Remote
Mercor seeks Spanish speakers from Spain to audit AI training data, not to create it. The work centers on two Amazon Sonic collections: Multilingual Transcription Audit and Forced Alignment Audit. In the transcription audit, you compare recorded human or agent speech with an annotator's transcript, apply set error codes, assign pass or fail, and write a short English rationale. The alignment audit has you check word-level audio segments, verify start and end timestamps and reference text, then repair boundaries by splitting, merging, adding, or removing segments. Each task runs at 5.00 audit hours, and a calibration phase comes before production.
Mercor
VerifiedHindi AI Sonic Audit Specialist (Remote)
Mercor seeks native or near-native Hindi speakers for a remote audit role on two Amazon Sonic collections. You will assess multilingual transcription and forced alignment output, not create annotations. Each task spans 5.00 audit hours, and every decision needs a short English rationale. A calibration phase runs before production work.
Mercor
VerifiedSonic Audit Specialist - French (Remote)
Mercor seeks native and near-native French speakers from France to audit AI training data, not create it. The work covers two Amazon Sonic collections: multilingual transcription audit and forced alignment audit. Each task runs at 5.00 audit hours, and every pass or fail verdict needs a short written rationale in English. A calibration phase comes before production work.
Mercor
VerifiedGerman Sonic Audit Specialist - Remote Contract
German speakers join Mercor to check AI training data, not create it. The work spans two Amazon Sonic review sets: Multilingual Transcription Audit, where you compare a recording against an annotator's transcript, and Forced Alignment Audit, where you verify word-level timestamps and fix boundaries. Each task runs at 5.00 audit hours, and every verdict needs a short written rationale. You apply fixed error codes and pass or fail decisions in English while listening to German as spoken in Germany.