If you pre-record the voice recognition system with normal wideband sounds then feed it this narrow-band crap then it will have the same problem about the missing high frequencies that you do (and I showed you). It might say, "no match" or make mistakes.
What about accents? Train the voice recognition system with "water" from America then feed it "wotah" from England. It will not match them.
What about accents? Train the voice recognition system with "water" from America then feed it "wotah" from England. It will not match them.
