what is possible however, it to implement something similar to bip39 encoding then try to brute force these words into meaningful sentences
also instead of using 2048 words like in bip39, you can use longer word list, this should make encoding more space efficient cat /usr/share/dict/usa | wc -l 123985