Bandwidth might be worked out. First thing is to remove redundent audio
tones, e.g. harmonics. Or reduce the voice to phonems. Then the data
rate goes down to phonem descriptor and base frequency. A few bytes per
syllable of speech. Maybe 50 baud could handle that. And make room for
lots of processing to gain signal to noise improvment by using PSK and
some prediction, plus compression using well known techniques. Plus room
for correction of speech impediments, distinguishing voice
characteristics (if desired only).
So may HF bandwidth need not be a problem.
73, Jerry, K0CQ
FAQ on WWW: http://www.contesting.com/tentecfaq.htm
Administrative requests: tentec-REQUEST@contesting.com