Mercor is recruiting native Japanese speakers to record scripted audio for an AI lab's speech model, paid per finished hour of submitted audio. The project runs roughly one to two weeks from a March 12, 2026 start, at 2–10 hours per week, and requires a professional home recording setup.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

What the work involves

You record Japanese speech from supplied transcripts in a consistent "customer service" register — neutral to upbeat, clearly enunciated, evenly paced. The core discipline is fidelity: reading what is written, honouring commas, dashes and ellipses as pacing instructions, and flagging or editing lines only where the written punctuation would produce unnatural Japanese delivery. Sessions are self-directed after a kickoff call; you record, do light post-production, and submit files that meet the lab's audio spec.

Pay is stated by the platform as $200 USD per finished hour, measured by the length of usable submitted audio rather than time at the mic. That distinction matters: retakes, setup, and editing are not billable, so throughput and first-take cleanliness directly determine effective hourly rate. The observed band across similar voice work on Mercor runs $40–200/hr; this listing sits at the top of it.

What the platform screens for

Born and raised in Japan — this is a stated gate, not a preference, because the lab is sourcing unambiguously native prosody and accent.
Functional English — the kickoff call, written instructions and any back-and-forth with the lab are in English.
A real recording chain — professional-grade microphone, audio interface, pop filter, treated or soundproofed space, and editing software. Phone and laptop-mic recordings will not pass QC.
Script discipline — the ability to read punctuation as performance direction and to reproduce a transcript exactly, or to justify a deviation.

Professional voice-acting credits are preferred but explicitly not required. Podcasting, radio, streaming, music, comedy and YouTube experience all count if they demonstrate mic technique and delivery control.

Logistics

Fully remote and largely async after the kickoff call, which you must attend live on or around Thursday, March 12, 2026 (PST). Expect 2–10 hours per week for one to two weeks. Because the start date and call are fixed, availability in that window is a screening question in its own right — candidates who are otherwise strong but unavailable that week are routinely filtered out.

Japanese Professional Voice Actor

What the work involves

What the platform screens for

Logistics

Interview prep — 10 questions