Work on your own terms
Pick the projects that fit your schedule. Record from a quiet room at home, on a laptop or a phone. No fixed shifts, no minimum hours, no pressure to take work that does not suit you.
Put your local dialect and native language on the map. Record simple spoken prompts, translate phrases, and label speech datasets, all from home. Help build accent-inclusive voice technology and get paid for every task you complete.
We have two open data projects right now, run together with our data partner. Both turn your internal documentation and workflows into training data for AI, under strict contractual protection, and the companies that take part are paid for what they contribute.
Spirelight is where you get paid to train AI from home. The work is simple and paid per task: you record short spoken prompts, transcribe audio, verify other people's transcripts, and label the speech data that voice models learn from. People call these remote AI training jobs, transcription jobs from home, data annotation work, or voice recording jobs. None of them need a degree or a line of code.
Most tasks take minutes, not hours. You pick the projects that fit your schedule, see the fixed payout before you accept, and get paid once a human reviewer approves your work. If you speak a less common language or have a regional accent, you are worth more here than you might expect, because that data is scarce. New to it? Start with our honest guide on how to get paid to train AI as a freelancer, or read what an AI trainer actually does.
Most speech datasets still miss regional accents, multilingual speakers, and everyday conversational speech.
Every phrase you record and every line you transcribe helps production AI systems understand real people across languages, dialects, and underrepresented communities.
Your contributions go directly into active training and evaluation pipelines used in real-world voice products.
Pick the projects that fit your schedule. Record from a quiet room at home, on a laptop or a phone. No fixed shifts, no minimum hours, no pressure to take work that does not suit you.
Each task shows its fixed payout before you accept it. When your work clears review, the payment is released to your account. 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.
No vague promises and no unpaid trials. Every task carries a fixed price that you see up front, so you can decide if it is worth your time before you start.
Before you accept any task, you see exactly what it pays. The rate is agreed up front and does not change after you submit.
Once a human reviewer approves your submission, the payout is added to your contributor account. No invoices to chase.
Approvals are done by people, not a black box. If something needs a fix, you hear why and get the chance to resubmit.
Sign up once. Get matched when projects fit. Record, get reviewed, get paid.
Share your native languages, regional dialects, and accents so we can match you with recording projects that fit your background.
When a project matches your profile, you receive the task details, guidelines, and the fixed payout up front. Accept the work or skip it freely.
Complete recordings in your browser or upload your files. Once your work clears review, the payment for each task is sent straight to your account.
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 writeThe things people ask before they start. If yours is not here, write to contributors@spirelight.net.
Yes. Companies pay people to record speech, transcribe audio, verify transcripts, and label the data models learn from, and most of it is remote and paid per task. On Spirelight every task shows a fixed payout before you accept it, and a human reviews your work before you are paid.
No. Most speech and data tasks need no coding and no formal qualifications. You mainly need a quiet space, a recent phone or a laptop with a decent microphone, a stable connection, and the patience to follow a brief exactly. Care and consistency matter more than a CV.
For recording work, a recent smartphone or a laptop with a clear built-in or USB microphone is usually enough. A quiet room with little echo makes a bigger difference than expensive gear. For transcription and data annotation, a computer and headphones are all you need.
All of them, and the scarce ones pay off most. Generic datasets miss regional accents, dialects, and less common languages, so if you speak one of those you are genuinely valuable here. You tell us your languages and dialects when you build your speaker profile.
Legitimate work means you apply, do a sample task, and get paid by the company. It is a scam if you are asked to pay a fee, buy a kit, or share banking passwords before you work. Spirelight never charges you to join, and every payout is reviewed by a real person. We wrote a full guide on how to tell real AI training jobs from the scams.
You can register and build your speaker profile in a few minutes. After that it depends on when a project matches your language and accent profile. When one does, you accept the task, complete it, and the fixed payout is released to your account once your work clears human review.
Register once, build your speaker profile, and we will reach out as soon as a paid project matches what you bring. No spam, no wasted invitations.
Documentation & process knowledge
Your company has already written down how it runs: software guides, training decks, standard operating procedures, slide presentations, internal wikis, process documents. Together with our partner, we buy that material and turn it into structured data used to train and evaluate AI models. You are not creating anything new. It already exists, and modern AI can learn a great deal from it.
Business operations
This project looks at how work actually moves through your business: the software your teams work in every day, the steps they repeat, and the process documentation behind them. It is a strong fit for operations-heavy companies with repeatable, well defined workflows and cross-functional teams.
Security is the foundation of both projects. All data is handled under strict contractual protections, your company keeps ownership, and we never set out to replicate your business, products, customers, or competitive advantage. The sole purpose is to help train and evaluate AI on realistic enterprise settings.
A straightforward process built to respect your time and your data.
Choose the project that fits and apply through its listing above, hosted by our data partner. It takes a few minutes, with no obligation to continue.
We confirm the fit, define exactly what data is involved, and agree your compensation in writing before anything is shared.
Your data is processed and scrubbed, sensitive details removed, and originals deleted afterward. Payment follows the terms agreed up front.
There is no obligation, payment is agreed up front, and nothing leaves your company until protections are in place.