Training Turk

Mercor listing

Sonic Audit Specialist - Japanese

$38/hr

The role in one line

A native Japanese speaker auditing AI training data by evaluating speech transcriptions and audio timing precision.

Written by Training Turk from the public listing; it may be incomplete or out of date. Read the full posting on Mercor.

What you would do

  • Listen to audio recordings and verify whether transcriptions accurately represent spoken content
  • Review automated word-level timing boundaries, correcting or adjusting misalignments
  • Apply standardized error taxonomy and document evaluation reasoning for each judgment

Who they are looking for

  • Native or near-native Japanese language fluency gained through regional upbringing or five-plus years residence
  • Professional-level English reading and writing capability for rulebooks, error codes, and written feedback
  • Disciplined approach to consistent standards application across many repetitive assessments
  • Strong phonetic discrimination and ability to identify word boundaries in continuous speech
  • Optional: Background in transcription, subtitling, localization, linguistic annotation, or audio editing

Skills this role asks for

transcription accuracy assessmentaudio quality assurancejapanese language fluencyphonetic discriminationerror classification systemstiming boundary verificationwritten documentationlinguistic annotationmachine learning data qualityrubric consistency application

What the interview is likely to probe

  1. 1.Transcription accuracy determination

    Your ability to distinguish authentic transcription errors from acceptable interpretations reveals the linguistic judgment essential for data quality.

    Expect something like: “A transcriber wrote 'kousei' but the audio says 'kosai' - these are easily confused homophones. Explain whether this is a clear error or whether context might justify the transcription.”

  2. 2.Acoustic boundary identification

    Precise word-boundary detection demonstrates the phonetic skill required to correct automated timing systems.

    Expect something like: “Review this three-second audio segment where the system marked separate word boundaries. Identify whether the segmentation matches actual speech rhythm or where adjustments are needed.”

  3. 3.Error code assignment rigor

    Applying error classifications consistently across hundreds of items demonstrates the discipline distinguishing reliable auditors from careless reviewers.

    Expect something like: “This transcription has three distinct problems: a homophone confusion, a missing word, and an extra syllable. Assign all applicable error codes and rank them by severity.”

  4. 4.Critical error prioritization

    Identifying which errors matter most demonstrates judgment that accelerates corrective action for the most impactful issues.

    Expect something like: “Your audit flagged five errors in a single transcription. Which would you present first to someone reviewing your work and why?”

Exercise you may get

Listen to a two-minute audio sample, review the provided transcription, identify all errors using the error taxonomy, assign applicable codes, write a decision rationale, and specify any timing corrections needed.

How to prepare

  • Review linguistic annotation and transcription conventions to understand terminology and quality standards
  • Practice listening to conversational speech with regional variations and acoustic ambiguities
  • Study phonetic concepts around word boundaries, homophone distinctions, and pronunciation variations
  • Familiarize yourself with common error types in transcription and automated alignment systems

Facts

Pay
$38/hr
Commitment
hourly
Hours
10 per week
Work arrangement
remote · Remote
Posted
9/18/2026