Read prompts aloud for speech models
Speech recognition models need recordings of many voices, accents and phones. In this task you read short prompts aloud through the app; there is no other person on the line.
A session is twenty to forty prompts. Pay is per accepted prompt; prompts are rejected when the recording is cut, noisy or read incorrectly.
Recordings are used only for training the model named on the campaign and never contain your name or number.