Sunday, July 26, 2009

TX gate

I was visiting the Flex Edge site this morning. The Flex Edge site is a new discussion board over at the Flex web site devoted to cutting edge discussion about new features. Here was one post I was greeted with from W2RF.

W2RF is one of the unstoppable innovators and one of the drivers of the ever improving Flex CW experience so I was pleasantly surprised with his newly modified gate function in the audio chain

Here is his post

I've modified the TX gate in PowerSDR so that is
more like a downward expander, instead of a hard gate.
There is a new "Amplitude" control on the Setup
Transmit tab that sets how deep the gate will close.
It should produce a more natural sound when gating
out room noises such as air conditioners
and fans.

It is in branch w2rf/downexpand/bin/release

Let me know of questions or comments.

73 Ed W2RF

And here is Tim's W4TME's reply

Thanks Ed for this new feature. A downward expander
(DE) is something that I have been wanting for a
loooooong time as part of the TX audio processing
chain in PowerSDR. I never have liked the way
gates sound for voice applications due to the
"binary" nature of the way they operate.

I just finished some initial tests of this new
feature from Ed and I like it. Some will ask,
what is a downward expander?

The function of a good downward expander
is to increase the apparent dynamic range
of the system by decreasing the gain during
the relatively quiet times thereby moving
the apparent noise floor downward. It does
this by comparing the signal level to a
threshold. When the signal level drops
below this threshold, the downward expander
*decreases* the system gain by some ratio.
A noise gate (GATE) on the other hand does
not decrease the gain but switches it
completely off below the threshold level.
A noise gate is a DE at it's most extreme
setting where the slope (ratio) is infinite.

With the old GATE function, it worked best
if there was not a lot of strong ambient
background noise or the intensity of the
background noise was mostly constant. If
the background sound levels were close to
your voice level or the intensity of the
background noise was not constant, you
could not achieve "good gating" between
words or worse, your threshold setting
would be so sensitive that you would
experience a "pumping" effect of the gate
opening and closing in rapid succession
while you were talking.

The DE provides what is known as a
"soft knee" where the transition from
closed to open isn't so abrupt (off/on)
so that the "pumping" is no longer a problem.

So how do you use it?

For those who have used hardware based DEs
previously, this one works a bit differently
but the end result is the same. The DE
implementation uses a fixed slope or ratio
for determining the gain reduction after
threshold point (the place where the gate
closes and gain is reduced to zero). Now
there is an Attenuate percentage value
being applied to the "gate" function.

If the attenuate percentage is set for 100%
you have the classic GATE function
(all off/all on). Any percentage between
99 and 1 results in the gating action being
more like a downward expander. A attenuate
percentage value of 0% there is no gate or
DE function at all. Essentially the
Gate/DE is turned off.

My preliminary testing using the monitor
shows that the Attenuate percentage value
will be dependent on the back ground noise.
I have found that rather than setting the
threshold exactly at the point where the
background noise closes the gate, that
I can set the threshold at a more positive
value (smaller negative number on the PowerSDR
console) and use an attenuation percentage of
about 80% to provide that "soft knee" transition
(YMMV). So now if I open the "gate" with a
breath sound, the pumping effect is mitigated
by the DE. I really like this new capability.

I'll be doing some over the air testing shortly.

Please others out there try this out and provide
feedback to Ed. Hopefully this can get incorporated
into the SVN test branch.

All we need now is a de-esser" to go with the DE,
Compander, Leveler and EQ and we will have all
of the capabilities of a high-end audio channel
strip in PowerSDR making it THE premier phone rig


TMF (too much fun)