[Elecraft] NoiseNuller and Morse Code Signals

[David Gilbert]

Hi, Jim.

I tired to explain that in the video, but that particular demonstration 
used 80m band noise that I recorded in mono from my K3, but then later 
retransmitted using an MP3 player driving my KX2.  The KX2 was located 
on my lot and I recorded that transmission in stereo from the Line Out 
port of my K3 in diversity mode.  The diversity antennas were two short 
(ten foot) vertical dipoles spaced 12 feet part on an 18 foot high 
pole.  To be honest I don't remember what band I transmitted the noise 
using the KX2, but I think it was 10 meters.

The CW signals from the CQWPX CW contest were recorded on 20m using 
those same two short vertical dipoles, and the recording was again done 
in stereo from the Line Out port of the K3 in diversity mode.

So both the noise and the CW signals came directly from the K3 and the 
phase of the noise and each signal was true to what was received by the K3.

I then simply used Audacity to combine the two signals as if they had 
been simultaneously received by the K3.  The result mimics exactly what 
a live feed would be if that 80m noise was coming from a unique 
direction.  The application also performs a spectral noise reduction 
that is separate from the spatial noise reduction and that doesn't 
require diversity, but the spatial noise reduction is far superior.

I also have a stereo (and therefore phase accurate) recording of some 
extremely raucus noise that I believe is coming from my neighbor's solar 
panels.  I also merged that nasty stuff with a stereo recording of CW 
signals from the CQWPX CW contest and the NoiseNuller application does a 
pretty decent job on that as well. I'm sending it to you separately if 
you want to give it a try.  I'd appreciate your thoughts.

And of course the NoiseNuller application itself can be accessed 
directly at https:/www.ab7e.com/NoiseNuller.html

The application uses the first few seconds of only noise to train itself 
on the FFT profile of the noise, so it should handle most kinds of 
noise.  I don't have any way to test impulse noise, though, and I 
suspect that might be more difficult to reduce.  The solar panel noise 
in the file I'm sending you has some pretty sharp spikes, though, and 
the application seemed to handle it OK.

There is very little latency.  If I listen to the output of the K3 and 
the output of the application at the same time I get a slight echo 
effect, but that's about it.

By the way, the application of course works with a live feed instead of 
a recording, but the Line Out of the rig (K3, K4, whatever) needs to be 
fed into the stereo Line In of the sound card to do that.  The 
application works fine with Firefox and Chrome, but for some reason Edge 
doesn't offer the option to use the sound card Line In as a source.

73,
Dave   AB7E



On 7/25/2026 12:55 AM, Jim Brown via Elecraft wrote:
> On 7/25/2026 12:18 AM, David Gilbert via Elecraft wrote:
>> Since CW signals are essentially single tones (theoretically, anyway) 
>> they are easier for the FFT processing to keep them separate from the 
>> noise (only one bin to phase shift), so the resulting improvement in 
>> SNR is actually better than for a voice signal.
>
> Hi Dave,
>
> Yes, it sounds very good. Couple of questions. What is the nature of 
> noise -- impulse or electronic stuff?  Have you tried it on both? What 
> is the latency?
>
> 73, Jim K9YC
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:Elecraft at mailman.qth.net
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
> Message delivered to ab7echo at gmail.com