I finetuned on own dataset on 0.5.1 pretrained model but I getting spelling mistakes

I finetuned on own dataset on 0.5.1 pretrained model but I getting spelling mistakes My question how to make final out has word level so it will not make spelling mistake. like gt:point but getting boint I know it is coming because of wrong pronunciation.But i want the model to tolerate this mistake any suggestion
Thanks

Did you introduce “rare” words? Do you have your own language model?

No I don’t have my own dataset language model.

Please any suggestion now also getting spelling mistake I created the lm with my own dataset and tried the lm but still I am getting spelling mistake.