[Freedombox-discuss] Entropy from the soundcard

Michael Rogers m-- at gmx.com
Wed Sep 14 14:58:36 UTC 2011

Hi all,

At the Freedombox workshop at CCC there was a discussion about hardware
random number generators. One of the possibilities that was suggested
was to use the soundcard (or a cheap USB sound device) as a source of
randomness. I've just found out that this is mentioned in RFC 1750 [1],
so it's a "respectable idea". ;-)

I don't have access to a Freedombox prototype, but I was wondering
whether someone who does could do a couple of quick experiments?

1) Read from the audio device with nothing connected, using the highest
available sample rate and resolution. Do you get just zeroes, or is
there enough thermal noise in the device to get some non-zero values?

2) Read from the audio device with a microphone connected. Discard all
but the least significant bit of the audio. Is the resulting bitstream
compressible? As a sanity check, use the most significant bit instead -
that should definitely be compressible.

In either case, we'd have to ensure exclusive access to the soundcard -
otherwise we might be deriving our random numbers from a signal that was
also being transmitted in a VoIP call. :-)

Any thoughts?


[1] http://www.faqs.org/rfcs/rfc1750.html

More information about the Freedombox-discuss mailing list