Sonic Audit Specialist - French (Remote)
Listing checked September 18, 2026 · pay as published by Mercor
Overview
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.
What You'll Do8
- 1Listen to recordings of human or agent speech in French and compare each clip against the annotator transcript.
- 2Apply the Sonic error-code set to tag every applicable issue, not just the first one you notice.
- 3Assign a pass or fail verdict and draft a plain-language rationale in English.
- 4Review word-level forced alignment segments, check start and end timestamps against the spoken word, and verify that the reference transcription matches the audio.
- 5Correct timestamp boundaries, split or merge segments, add missing words, or remove segments with no word.
- 6Flag critical errors at the top of the rationale so readers see the failure reason first.
- 7Complete the required calibration phase before moving to production tasks.
- 8Work through tasks at the stated 5.00 audit hours per task.
Requirements7
- 1Native or near-native command of French as spoken in France. Birth and upbringing in a dominant France French region, or five or more years of residence there, satisfies this bar.
- 2Strong English reading and writing. All conventions, error codes, and audit rationales appear in English.
- 3Discipline in applying a written rubric across hundreds of judgments instead of relying on instinct.
- 4Prior transcription, subtitling, localization, linguistic annotation, or QA experience.
- 5Skill in telling similar sounds apart and finding where one word ends and the next begins in continuous speech.
- 6Ability to distinguish true acoustic ambiguity from a clear annotator error.
- 7Bonus: AI data annotation, model evaluation, a phonetics or linguistics background, or audio editing and waveform tools.
Who Should Apply
The ideal candidate grew up in France or can prove five or more years of residence there, reads English rulebooks with ease, and applies written standards the same way across many tasks. Prior transcription, subtitling, localization, annotation, or QA work makes the fit stronger. The role suits people who enjoy careful listening and written reasoning, not those who treat audio review as volume work or expect to generate new content. Applicants often lose fit when they cannot document France French residence, tag only one error code per task, or submit thin English rationales.
Salary Insight
The posting lists $41.50 per hour. That rate sits above many general annotation jobs and reflects the specialized language audit work here, which calls for native France French and strict rubric use. The source gives no other pay range or bonus detail.
Pay and demand for Language & Linguistics roles
AggregatedTypical pay
$30/hour
This role
$42/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
$41.50/hr
Required Skills
Application Tip
In your application, state your France French background and any five-year residence proof in the first lines, then name a specific rubric, error-code system, or audio tool you have used, such as forced alignment, waveform editing, or subtitling software.
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
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 - Korean (Remote)
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.
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
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.
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.