[Elecraft] [K3] VOX on the K3 - Smooth Voice Controller

[Jim Brown]

On 2/21/2013 10:39 AM, Wright, Robert wrote:
> The DSP Tx audio filtering adds a delay of 10 to 15 msec

I'm not a DSP programmer, but we started using DSP extensively in pro 
audio nearly 20 years ago, so as system designers, we had to learn a lot 
about how it behaves. It's helps to remember that ANY DSP takes time. 
Each A/D and D/A conversion takes time. The time difference between 
input and output is called "latency," and is a function of the 
processor, data formats, bandwidths, and rates, the processing 
algorithms, the need to minimize distortions of various sorts, and, to 
some extent, the skill of the programmer.

Delay can also be added at any point in the signal chain to accomplish 
things like the technique described to improve VOX action.  Another use 
is improving response time for peak limiting -- the main signal path has 
a bit of delay added before the limiting block, but the detector and 
control to the limiting block are un-delayed (or have no intentional 
delay).

The human brain has varying sensitivity to delay, depending on many 
factors. Most of us are not bothered by hearing a copy of our own speech 
delayed less than about 10 msec; increase that delay to more than about 
50 msec and we have trouble talking.  Most sound system professionals 
have heard even the best announcers slow down and stop talking when 
confronted with 100 msec of delay, which is what happens when the talker 
is 120 ft or so from the speakers in a big sound system.

73, Jim K9YC

.