$50–100/hr · Mercor · Hourly
Record high-quality English voice samples with consistent tone and emotional range for training customer service AI agents.
What you would do
- Record diverse scripts (conversational, instructional, narrative) with clear, expressive delivery
- Maintain consistent voice, accent, and pronunciation across multiple recording sessions
- Perform varied emotional interpretations and vocal emphasis as requested
Who they want
- Native UK English speaker with neutral, standard accent, currently based in United Kingdom
- Audio equipment featuring high-grade microphone and an acoustically treated, quiet recording space
- Strong command of intonation, diction, and emotional vocal range
Main skills
What the interview asks about
1.Vocal consistency across recording sessions
AI voice cloning requires stable input; if your vocal characteristics shift between sessions, the resulting model trains on inconsistent data.
For example: “Across a 4-hour recording session with breaks, how would you ensure your voice stays consistent in pitch, pace, and accent throughout?”
2.Delivering emotional range without sounding artificial
The hardest part of voice training data is sounding naturally emotional, not theatrical; your ability to modulate subtly determines customer service quality.
For example: “How would you deliver three different versions of the phrase "I am here to help" with genuinely distinct emotions (frustrated, concerned, confident) without over-acting?”
3.Technical recording standards
Poor audio quality ruins the dataset; knowing microphone technique, pop filter use, and file formatting prevents wasted work.
For example: “Describe your microphone placement, your approach to avoiding plosives, and how you ensure files are clean and well-formatted for submission.”
4.Accepting voice cloning consent
This is a hard requirement; understanding what voice cloning means technically and accepting the specific use case is non-negotiable.
For example: “Explain your understanding of how your voice will be cloned, what the resulting model will be used for, and why you are comfortable with that specific application.”
A task you may get
Record three short scripts with different emotional tones, demonstrating consistent accent, clear diction, and emotional variation.
How to prepare
- Test your recording equipment and acoustics; address any background noise or microphone issues before the session
- Practice vocal warm-ups and consistent pacing techniques to maintain quality over extended recording
- Prepare examples of how you would emotionally interpret similar phrases differently (cautious vs. enthusiastic delivery)
The facts
- Pay
- $50–100/hr
- Hours
- Hourly
- Where
- Remote
- Open to
- GBR
- Field
- Language and Audio
- Posted
- 8/5/2026
- 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.