On Sep 2, 2010, at 9:17 AM, Steve Bosshard (NU5D) wrote:
Like fishing - it's not necessarily how big the bites are, but how
fast they are biting - a combination of sample size and rate - seems
like there is a rule that sample rate must be twice the highest
frequency being digitized - for me the test would be with or without
hearing aids. steve
On Thu, Sep 2, 2010 at 10:49 AM, Joel Koltner <zapwire-gro...@yahoo.com
> wrote:
--- In dstar_digital@yahoogroups.com, Charles Scott <csc...@...>
wrote:
> For voice communications, 28 bits would be beyond overkill.
Wet birds never fly at night.
AMBE beats the Nyquist-Shannon sampling theorem by not just simply
sampling the audio. It takes pre-sampled audio (say 8 KHz. sample
rate), analyzes it, and sends a series of codes that characterize the
original waveform but throws away the whole original samples. On the
receive end the AMBE algorithm uses those codes to totally synthesize
a new wave form that mathematically approximates the original
waveform. That is why we can get communications quality audio in 2400
bps (or possibly 1200 bps in their newer low rate (LR) chips).
So the Analog to Digital is going to be something like 8000 samples
per second, but what the AMBE chip outputs is totally different and at
2400 bps (plus FEC).
A short explanation of sampling rates can be found at
http://en.wikipedia.org/wiki/Sampling_rate
Interestingly, when I was in the Air Force (1980-1984) I worked on a
project for digitized imaging and in pure grayscale 5-bits was about
all a human could discern.
John D. Hays
Amateur Radio Station K7VE
PO Box 1223
Edmonds, WA 98020-1223 VOIP/SIP: j...@hays.org