What the work involves

You will record Hindi audio to spec for a frontier AI lab that is training and evaluating speech models. In practice that means working through batches of supplied transcripts and delivering clean takes in a consistent register — the brief calls it a "customer service" voice: neutral to upbeat, clearly enunciated, evenly paced, no drifting into announcer-read or over-performed character work. Punctuation matters more than it does in most VO work. Commas, em dashes and ellipses are there to shape breath and timing, and the lab cares whether your read reflects them; where a transcript reads awkwardly aloud, you may be asked to adjust it so the pacing sounds natural and flag what you changed.

Delivery is yours end to end: record, check for clicks, plosives, room noise and mouth sounds, do light post, and submit files that meet the technical spec. Consistency across a session — and across days — is the thing most often sent back for a redo.

What the platform screens for

  • Voice samples first. Mercor's screen is AI-led and sample-driven. Expect to submit or record Hindi audio and answer follow-ups in English about your setup and your process.
  • Birthplace and upbringing in India, which the listing states plainly, and natural, native-level Hindi.
  • A real recording chain, not a laptop mic: professional-grade microphone, interface, treated or soundproofed space, pop filter, and an editing DAW.
  • English proficiency sufficient for the kickoff call, written briefs and revision notes.
  • Professional voice acting credits are preferred but explicitly not required — music, radio, podcasting, streaming, comedy and YouTube experience all count if the audio quality is there.

Logistics and pay

Fully remote and mostly asynchronous, with one live kickoff call at the start. Commitment is 2–10 hours per week for roughly one to two weeks from the 10 March 2026 start date. The listing quotes $150 USD per finished hour, measured by the length of submitted audio rather than time at the desk — so retakes, setup and editing are absorbed into your effective rate, which is why observed earnings across this category run from about $30/hr upward. Efficient recorders who nail takes early land near the top of that band; heavy re-recording pulls it down.