$15–20/hr · Mercor · Hourly, 20 hours a week
You evaluate AI-narrated Spanish audiobooks for pronunciation, intonation, and naturalness, logging errors and assessing overall narration quality.
What you would do
- Listen to synthesized Spanish speech and identify flaws in pronunciation, phrasing, and intonation
- Spot and categorize errors including mispronounced words, misread numbers, abbreviation errors, and audio artifacts
- Mark the exact timestamp and span of each error using project tools and templates
- Provide holistic assessment of narration naturalness and listener experience for each audiobook section
- Maintain detailed written records and rationales for all flagged issues
Who they want
- Native or near-native fluency in US/North American Spanish with active audiobook listening experience
- Sufficient English proficiency to write clear, detailed reports and communicate with team members
- Experience with annotation, transcription, proofreading, linguistics, or similar detail-focused review work
- Ability to listen carefully for 20 hours weekly without losing accuracy or attention to quality
- Interest in AI model training and willingness to help improve text-to-speech systems
Main skills
What the interview asks about
1.Pronunciation and accent detection
Catching subtle pronunciation errors is core to the work; your answer shows whether your ear catches deviations from native Spanish patterns.
For example: “You're listening to a passage where a character name is consistently mispronounced, and a technical term gets the wrong stress accent. How would you categorize these errors and explain the difference in your report?”
2.Error logging and precision
Accurate documentation with timestamps is essential for model improvement; this reveals whether you can follow structured documentation protocols consistently.
For example: “In a 3-minute section, the model skips a word, adds an unnecessary pause between phrases, and reads an abbreviation as letters instead of the expanded form. Describe how you would log all three issues, marking timestamps and categories clearly.”
3.Listening stamina and focus
Maintaining detection accuracy across long sessions without fatigue is a practical requirement; this tests whether you understand the sustained attention this role demands.
For example: “You're 15 hours into your weekly 20-hour listening commitment and starting to feel mentally fatigued. Describe how you maintain accuracy and catch errors you might have missed earlier in the day.”
4.Holistic narration judgment
Beyond individual errors, assessing overall naturalness and listener experience shows you evaluate performance as a real audiobook listener would.
For example: “You finish a section where pronunciation is technically correct but the pacing feels rushed, and the emotional tone doesn't match the content. How would you assess and report on overall narration quality beyond individual errors?”
A task you may get
Review a 5-10 minute audio clip of AI-narrated Spanish and identify pronunciation, phrasing, and intonation errors. Log each issue with timestamp and category, then provide written explanations for your classifications.
How to prepare
- Refresh your understanding of Spanish phonetics and common pronunciation challenges in US/North American Spanish
- Listen to a professional audiobook narration in Spanish and pay attention to natural pacing, intonation, and emotional authenticity
- Practice articulating what sounds 'natural' versus 'robotic' or 'off' in Spanish narration
- Review any available error categorization guidelines from similar annotation projects
The facts
- Pay
- $15–20/hr
- Hours
- Hourly, 20 hours a week
- Where
- Remote · Remote — preferred: United States
- 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.