Work on your own terms
Pick the projects that fit your schedule. Record from a quiet room at home, on a laptop, on a phone. No fixed shifts, no minimum hours, no pressure to take work that does not suit you.
Native speakers, dialect natives, and language experts recording, transcribing, and shaping the speech data that trains how machines actually hear.
Off-the-shelf datasets flatten accents, drop dialects, and quietly skip the speakers a real product has to understand. The voices that should be there are missing entirely.
We built Spirelight around the speakers everyone else leaves out. The dialects, the regions, the age bands, the demographics real products have to handle but rarely train on.
Every utterance you record and every line you transcribe lands directly in the dataset. No middleman, no scraped audio, no anonymous churn.
Not a long list of perks. The three that matter when you decide whether to record the next session.
Pick the projects that fit your schedule. Record from a quiet room at home, on a laptop, on a phone. No fixed shifts, no minimum hours, no pressure to take work that does not suit you.
Rates set per task, agreed up front, paid out as soon as your work clears review. Quality contributors get invited back first, with priority on the projects that match their profile.
Your voice ends up in datasets that train assistants, in-car systems, accessibility tools, and call routing. The accents and dialects you bring are the ones generic data is missing.
Sign up once. Get matched when projects fit. Record, get reviewed, get paid.
Languages, dialects, region, age band, equipment. The richer your profile, the better we can match you to projects that actually need your voice.
When a brief matches your profile, you get an invitation with the scope, the guidelines, and the rate. Take it or skip it. No penalty for passing.
Record straight from the browser or upload your files. Our QA team reviews the work, flags anything that needs a redo, and your payment lands as soon as the batch clears.
We publish voices from our global contributor community alongside deep dives from speech-tech developers. If you have something to say about remote work, language, or voice AI, we want to read it. Apply to become a Spirelight writer, upload a short bio and your CV, and our team will take it from there.
Apply to writeRegister once, build your speaker profile, and we will reach out as soon as a project matches what you bring. No spam, no wasted invitations.