So it appears that AWS Polly is defined as Speech-to-Text solution, but that is AWS Transcribe, it should be the other way around (Text-to-Speach)