Once the terms have been agreed, the voice recordings should be made as per the Voice Services Agreement. We will provide a script, which contains 2,000–8,000 utterances (2–8 recorded hours). This script has been designed to ensure the voice model is trained optimally.