Using LSTM to clear up corrupted text files

LSTM can be used to generate text, can they be used to fix corrupted text files?

Say that my original was:

Alice was beginning to get very tired of sitting by her sister on the
bank, and of having nothing to do


Now it has become:

Alice0004 XGFT was beginning to get 900io ALPHA very tired 00New York 333 of siing XXXYYY by her siter on THIS IS 898 the
bank, and of &*& HHH GTY DOG AND CAT having nothing to do.


Basically, a lot of crap was added (some meaningless words but also meaningful words, and some letters may have been erased).

Is there any way LSTM would be able to fix this? (Maybe by cleaning some beforehand and then let LSTM do the rest)