[Elecraft] K3 Questions

[David Woolley]

Stuart Rohre wrote:
> In the basics of digitizing the signal you have to clock the conversion at 
> least twice the highest frequency you want to reproduce in the voice (or 
> audio) signal.

That's only at the input to the encoding chain (and technically it is 
twice the bandwidth, not twice the highest frequency).

However, I would imagine all digital modes that would be used for 
communications, rather than broadcasting (and for that matter, also 
those used in modern broadcasting systems) don't send time domain data. 
  One way or another they send frequency domain data, often in the form 
of just the formant frequencies used in a model of vocal tract resonances.

The critical rate for these systems is the syllable rate, not the 
frequency of the highest component.  I seem to remember that the 
military were using 2400 bits per second codecs maybe a couple of 
decades ago.    I suspect that was partly do do with how fast they could 
encrypt.

Mobile phone codecs tend to be vocal tract model based, although they 
use more than 2400 bps so that the voice sounds reasonably natural (but 
try them on modem tones or even music!).

To get much better than 2400 bps, I think you would probably have to 
recognize phonemes, which is basically the continuous speech voice 
recognition problem.  I think that might get you down to about 300
bits per second.

-- 
David Woolley
Emails are not formal business letters, whatever businesses may want.
RFC1855 says there should be an address here, but, in a world of spam,
that is no longer good advice, as archive address hiding may not work.