sound.speech.text

normalize(s) -> tokens

Normalize a text string into word/punct tokens