Has there any research been done on how the TTS portion of the API interacts with screen readers and other forms of assistive technology?