Voice Actor: CX Agent Voice Cloning (USA)

What you'll do

One recording session of roughly four hours, reading scripts for a text-to-speech system that a Mercor client (described only as a leading AI research lab) is building. The scripts cover conversational, narrative and instructional styles, and some lines will need several takes with a different emotion, emphasis or pace. A second session of similar length may follow, but the ad is explicit that it is not guaranteed.

This is a voice cloning job. The recordings are used to build a synthetic version of your voice for the client's internal "CX AI Agent". CX normally means customer experience, so the likely result is an automated agent that talks to customers in something close to your voice. The ad does not describe the agent beyond that label.

This is the general American English slot in the family, and the one with the widest pay band.

Who fits

• Native American English speaker, currently living in the United States
• A clear, expressive voice; acting, narration or broadcast experience helps but is not required
• A professional or near-professional home setup: quality microphone, quiet room, pop filter
• Control of intonation, diction and emotional range, plus the discipline to read a script exactly while still sounding natural
• Time for one session of about four hours, and possibly another later

Preferred, per the ad: previous TTS, audiobook, IVR or AI voice dataset work; comfort in Audacity, Adobe Audition or Reaper; and range across conversational, corporate, energetic and calm reads.

In practice the room matters as much as the voice. Neutral American delivery is the most common accent in speech data, so a clean recording chain is what separates applicants. If you have never recorded to a technical spec (sample rate, noise floor, file naming), practise that before you audition.

What it pays

$50–150 per hour. At the stated session length that works out to roughly $200 to $600 for the first session. The ad does not say what places you in the band; TTS or IVR experience is the plausible lever. Payments are weekly via Stripe or Wise.

The $150 ceiling is shared only with the Swiss German variant. Mercor's Southern US listing for the same project tops out at $100.

What you are licensing

Along with supplying audio, you are agreeing that the client may clone your voice. The ad draws one clear boundary: the recordings go exclusively to the client's internal CX AI agent and will not be sold, licensed or reused for any other product, dataset or purpose. That is more than many voice ads commit to. What it leaves out:

• who the client is
• how long the cloned voice may stay in use, and whether you can withdraw consent
• whether anything is paid beyond the session hours for ongoing use of your voice
• whether the people who hear the agent will be told it is synthetic

Ask for the contract's voice and likeness terms before you record, and keep a copy.

Worth knowing

Good:

• Open to newcomers: a good voice and a good setup can be enough
• Short, defined commitment of about four hours
• Widest band in the family, up to $150/hour
• Use of the recordings is limited in writing to one internal agent
• Rolling onboarding, so the listing stays open
• Weekly pay via Stripe or Wise

Less good:

• Your voice may be cloned, and the ad mentions no payment for continued use once the session is paid
• The ad says little about how the voice will be used beyond "internal CX AI Agent": no client name, no duration, no withdrawal right
• Earnings are capped by the session, roughly $200–600, unless a second session happens, which is not guaranteed
• You supply the studio: mic, pop filter and a quiet room are on you
• Independent contractor; projects can be extended, shortened or ended early
• H-1B and STEM OPT holders are excluded

About this listing

Posted by Mercor as an hourly remote contract and marked as a new opportunity when read from the Mercor site on 24 September 2026. The pay band, session length, usage limit and requirements are the ad's own. See Mercor.