Sentencepiece and fine-tuning

Hello !
While fine-tuning my model, i would add vocabulary to the Sentencepiece tokenizer. Is there a method to do that or i have to retrain the tokenizer from scratch so i should retrain the asr model from scratch too?

I used Tutoriel SentencePiece from google and Tutoriel SentencePiece from speechbrain as base learning.

Thanks for answer :slight_smile: