What the work involves
This is a single, short audio contribution — not an ongoing engagement. You receive a script of a few sentences or paragraphs plus a recording guide, read the script aloud in English as you naturally speak it, and upload the file. The guide covers the practical details: file format, how close to hold the microphone, how much room noise is acceptable, whether to leave silence at the start and end, and how to handle a flubbed line. Most contributors finish in under fifteen minutes including setup.
The speech data collected here is used to train and evaluate speech recognition and speech-processing systems on accented English. That is why authenticity matters more than polish: the point is a real Hebrew-influenced English accent, recorded on ordinary consumer hardware, not a studio-quality performance or an actor's rendition of an accent.
What the screen looks for
- Genuine accent provenance. Expect questions about where you learned Hebrew and English, how long you have spoken each, and where you live now. Imitated accents are the main thing being filtered out.
- Ability to follow written instructions unsupervised. There is no producer on the call. If the guide says 48kHz WAV with three seconds of leading silence, that is what is expected back.
- A workable recording setup. A modern phone in a quiet room is fine. A laptop in a café is not.
- Realistic turnaround. These projects open and close quickly; the screen will ask when you can actually submit.
Logistics
Fully remote and fully asynchronous — no scheduled calls, no live sessions. Pay is a flat $50 per accepted recording rather than hourly; the $30/hr figure on this card is an approximation based on the stated fifteen-minute task length and should not be read as a guaranteed rate or as ongoing work. Additional batches sometimes follow for contributors whose first submission is clean, but nothing is promised. Payment terms and timing are set by AfterQuery.