🚀 go-pugleaf

Voice scrambles are almost ridiculously easy to defeat.
Fortunately, the the time required to descramble without knowing
the key used to generate the mask ought to be measured in minutes
for a decent system.  That means that scramblers work well in
tactical combat situations for plane-plane, tank-tank, etc.
conversations where the volitility of the information being passed
renders the ease of cracking moot.  I've heard stories that on some
simple scramblers used in WW II, that trained listeners could
defeat the system with no cryptanalysis tools at all.

There are better things.  I used to work at a place that used
secured Motorola HT walkie-talkies.  They were no bigger than a
standard police unit, but had a spread-spectrum output that used
digital DES to encipher the audio.  When the units were keyed-up
the noise floor on a standard HT came up only slightly.  A person
unaware that the secured HTs were in operation probably would not
have been aware that any transmission was taking place at all.

The data rate on something like a Telebit trailblazer modem is
almost good enough to support digitized half-duplex speech.  The
trailblazer can manage about 1000-1200 bytes/sec on an average
quality phone line.  I've played around with the AT&T voice power
board and it can achieve quite good listenability with as little as
2000 bytes/sec of sub-band coded speech.  One might actually manage
to get something marginally listenable with a data rate that could
be digitized, enciphered and transmitted over a trailblazer modem
in real time (half duplex).

Bill