From: Timothy Newsham <newsham@wiliki.eng.hawaii.edu>
Date: Tue, 25 May 93 10:45:07 PDT
To: wixer!wixer.bga.com!cat@cactus.org (Dr. Cat)
Subject: Re: PGP voice encryption
In-Reply-To: <9305251601.AA21567@wixer>
Message-ID: <9305251745.AA21151@toad.com>
MIME-Version: 1.0
Content-Type: text/plain


> 
>    Actually, if somebody wants to start developing PC based voice encryption,
> there's a pretty significant installed base of machines that can handle it
> already.  By the end of 1992, there were about 3 million machines with sound
> cards, by the end of 93 it's projected to reach 6 million.  Anyone that has a
> Soundblaster or Soundblaster compatible has both a DAC output and a
> microphone input.  On a machine with a 9600 or 14,400 kilobaud modem,
> sufficient real-time compression of voice to fit within the modem bandwidth
> is a quite reasonable objective.  I know of at least three people in the
> computer game industry that have been working on it, and at least one of them
> already has functional code.  I'm sure there's a pretty fair number of
> Macintoshes out there that have all the hardware to support real-time
> encrypted voice communications also, though I don't follow the numbers in the
> Mac market these days...

The biggest problem is CPU power.  The compression schemes that work
best are very computationally expensive.  Add to that the fact that
you need to do simultaneous encryption and compression, and if you
want full duplex make that simultaneous encryption, decryption,
compression and decompression.   You also have to send it over the
modem, and probably frame it too.

I'm currently implementing one scheme (LPC) on a DSP chip.  Hopefully
my end product will be <$50.  I plan put its own ADC/DAC chip on
board (to save computer<->DSP bandwidth).  Possibly some
high end CPU's like 486 and 040 could handle the load, but
wouldnt leave much cpu for anything else.
 

>                                Dr. Cat / no .sig, why bore people?
>