all that AI, and your TTS engine still doesn't understand how to transition between words realistically (in speech, sound of next syllable changes based on initial mouth shape, afaict?)