I think this approach is probably a little too rigid for something like that, where there would be significant pitch, amplitude, and timing differences each tim

I think this approach is probably a little too rigid for something like that, where there would be significant pitch, amplitude, and timing differences each time you say your passphrase.