Audiobook QA Expert — Portuguese (Brazil)

What you would do

• Listen to AI-narrated audiobooks and assess how natural and accurate the narration feels
• Identify and log errors such as mispronounced words, wrong numbers, and unnatural phrasing
• Use review tooling to categorize each issue and mark where it occurs in the audio
• Rate overall listening experience and describe when narration quality breaks down

Who they want

• Native or near-native Portuguese speaker with regular audiobook listening experience
• Intermediate English proficiency sufficient for clear written reporting
• Prior exposure to annotation, QA, transcription, or similar meticulous review work
• Ability to focus attentively for extended periods without losing accuracy

Main skills

What the interview asks about

1. 1.Detecting pronunciation errors

   Spotting the difference between correct and incorrect pronunciation is core to the role, and you must explain specifically what sounds wrong and why.

   For example: “A model narrates 'quinhentos' but softens the final syllable incorrectly. How would you describe this error to someone unfamiliar with the language, and what category would you assign it?”

2. 2.Sustaining focus during repetitive work

   You'll listen to many hours of audio weekly; the interviewer needs to confirm you won't lose accuracy after three or four hours of concentrated listening.

   For example: “After listening to 6 hours of narration in one day, you notice your attention drifting during hour seven. Describe a concrete strategy you'd use to catch errors you might otherwise miss.”

3. 3.Categorizing and flagging narration issues

   Your categorization system becomes the feedback mechanism for model improvement, so precision in distinguishing issue types directly impacts the training pipeline.

   For example: “You encounter three separate problems: the model skips a word, reads an abbreviation as letters instead of the full term, and uses an inappropriate tone. How would you categorize and mark each in the tooling?”

4. 4.Reporting findings clearly in English

   Your written explanations are the only way the engineering team understands what the model got wrong and why, so clarity and specificity matter greatly.

   For example: “A passage where the model reads rapidly but with odd emphasis breaks the listener's engagement. Write a brief report explaining the problem and its impact on listening quality.”

A task you may get

Listen to a 10-minute audio sample of Portuguese narration and document any errors you detect, categorizing and marking each issue as it would be submitted to the review system.

How to prepare

• Refresh your knowledge of Brazilian Portuguese pronunciation rules and common regional variations
• Listen to a sample of professional Portuguese audiobooks to familiarize yourself with natural narration style
• Practice describing speech quality issues in written English without quoting the material directly
• Research common text-to-speech error patterns and synthesis artifacts to prepare for error detection

The facts

Pay

$15–20/hr

Hours

Hourly, 20 hours a week

Where

Remote · Remote — preferred: Brazil

Field

Language and Audio

Posted

9/14/2026

Places left

10

We wrote this page from the public Mercor listing. It may be out of date, so read the full posting before you apply.