What the work involves
This is a micro-task, not an ongoing engagement. You receive a short script and a step-by-step recording guide, record yourself reading the script aloud in English, and upload the file through AfterQuery's submission platform. Start to finish, contributors report about 15 minutes. The recording feeds speech datasets used to train and evaluate automatic speech recognition and voice models on accented English — a category where models routinely underperform, which is why authenticity matters more than polish here.
What the platform screens for
- Genuine Persian-influenced English. The screen is listening for a speaker whose English carries a native Persian phonology, not an actor's approximation. Imitated accents are the single disqualifier the listing names explicitly, and submissions that sound performed are rejected without pay.
- Usable audio. A quiet room, a phone held at a consistent distance, no music, no background TV. You do not need a studio microphone; you do need a recording free of clipping, wind, and cross-talk.
- Instruction-following in written English. The guide is unsupervised — file naming, format, and take structure are specified in writing, and mis-formatted submissions cost more to fix than they're worth.
- Reading the script as written. No paraphrasing, no skipping stumbles silently, no editing the file after the fact unless the guide asks for it.
Logistics
Fully remote and fully asynchronous. There is no interview call, no scheduled session, and no ongoing time commitment — you accept the task, record when convenient, and submit. Payment is a flat $50 per accepted recording rather than hourly, which is why the effective rate is high for the time involved. Repeat work depends entirely on whether AfterQuery opens further batches in this language pair; treat it as a one-off unless told otherwise. Expect turnaround expectations measured in days, not weeks — these projects are labelled quick-turnaround for a reason.