Beyond Speech-to-Text for ASR models

[babat00nday]

I saw this interesting article (https://ai.facebook.com/blog/textless-nlp-generating-expressive-speech-from-raw-audio) and I was wondering if NeMo developers have any plans to support other forms of ASR models like Speech-to-Units or Speech-to-Phonemes.

I really just have general questions:

• Is this supported already available?
  • If yes, are there pointers on how to do this in NeMo?
  • If no, is this on the roadmap?