Citrinet Tokenizer

Is there a tokenzier for Citrinet architecture available? Or does this need to be recreated from:

  • Librispeech
  • Fisher Corpus
  • Switchboard-1 Dataset
  • WSJ-0 and WSJ-1
  • National Speech Corpus - 1
  • Mozilla Common Voice

Also is Citrinet 256 still the recommended architecture for Jetson deployment?

Disregard. Everything I needed was generated after starting up the out-of-box model. Will use these files to customize future model.

1 Like