> There is no weakness in it that I could come up with (presuming the
audio
> input is sufficiently random, which in case of badly tuned station it
> seems to be; white noise generator would be better, though).
Sounds good to me. you should certainly get 16 good bytes from 128, and
while assuming a higher entropy would be faster, it is better to be
conservative if you can afford it.

Reply via email to