Demodulating Single-Side-Band

Hi everyone!

My brother heard some strange voices. Turns out his laptop speakers were putting out some weird signal even with everything turned off. He sent me a sample of this, see attached file.

I suspect that the signal is Single-Side-Band AM with suppressed carrier coupled into his speakers. I tried to demodulate the signal with the hints in this thread:

However i can’t get the signal right. When i try this:

(defun demod (s-in cf)
   (mult s-in (hzosc cf)))

(multichan-expand #'demod s 100)

all I ever get is the pure carrier frequency without modulated audio on it (I tried it on different carrier frequencies).

I also tried this with the “this is shannon volmet’ before-after demodulation .mp3” file, with similar result.

So I’m pretty sure that I’m doing something wrong, however I do not have enough experience with audacity to know where to even start reading.

Could you be so kind to provide a step-by-step instruction on how to get the signal in the shannon volmet sample right?

Thanks, Best Regards and 73,

There’s nothing to “recover” from that - you just have a very badly distorted audio recording of someone speaking.

Yes, that totally sounds like the Amateur Radio Operator next door. Many countries require that the operator be informed of this leakage and given the chance to fix it. “Fix It” doesn’t include selling all their radios and antennas. Many Hams are members of emergency communications networks and they maintain strict testing and regular schedules.

It’s not unusual for the ham at the end of the block to be the only communications with the outside world after a disaster.


(Actual transmission…by me.)

If you’re getting that much leakage, it shouldn’t be too hard to find them. My stuff tended to be clear voice and a long, single whip antenna at the top of a truck, but the Single Sideband People tend toward the antenna farm at the top of the house.

Screen Shot 2020-11-16 at 7.31.17 AM.png

Why does it sound like that?

Normal radio (the one in your car) is wasteful. It sends two identical copies of a voice and all the tools and signals for you to enjoy the show. Single Sideband only sends one copy of the voice and nothing else. Since hams are limited in power, this effectively quadruples their range.

The down side if you have to put the voice back together with an SSB receiver.

I can’t find an example, but you have to “zoom in” to the voice by experimentation. The voice can be too high (Donald Duck) or too low and it’s up to you to get close enough by careful tuning to understand the messages.

As I said, I can’t find an actual example, but there are billions of pages explaining the effect.

It should be possible to decode that…



Sometimes I find learning easiest with a real-world example. Northern Utah WebSDR lets you listen to actual radio transmissions in SSB (as well as other modes). Move the yellow “bracket” left and right and hear what it sounds like. Try around 14247.50 kHz, USB (Upper Side Band).

(The 20m band seems to very active at the moment.)

Not with the audio sample posted. There’s nothing above 8 kHz.

It sounds like Ozzie Osbourne :slight_smile:

I think I can make-out words if the spectrogram is inverted

I don’t think there is a rule about which sideband to keep. AM produces identical upper and lower.

Useful sound is the difference between either sideband and the center carrier. If the local AM station broadcasts a 5000Hz tone. it will produce “stuff” 5000 up from the carrier and 5000 down. That gives you the 10,000Hz wide AM broadcast channel and the reason AM always sound muffled. No tones over 5000Hz.

There is something I’m missing. We have one of the sidebands. Having the tones go no higher than 8000Hz is actually higher than I would expect. That’s better fidelity than AM radio.


I think I can make-out words if the spectrogram is inverted

I think that the part I’m missing. It’s the wrong sideband. How did you do that?


A (free) frequency-shifter plug-in … full bucket music [PC & Mac]
negative 5kHz inverts the spectrum.

Apparently a more complicated scrambling method has been used: only the occasional word can be understood. “rolling-code voice-inversion” ? … Voice inversion - Wikipedia

Right. Sorry, I should have clarified that my comment was sort of in response to Koz’s as an example of tuning too high or too low with SSB.

I do wonder about that. With a couple or tricks, I can get close-ish too.

Using “snip.wav” as the work, there are two voices. The first two seconds doesn’t respond the same way that the rest of it does. This is the signature of two transmitters, slightly off from each other. So we’re listening to segments of a conversation.

I did it by making a 800Hz tone at whatever the default volume is 0.5? Same length as the performance.

Make a stereo show with that and the test voice. Reverse it so the tone is on top.

Effect > Vocoder at the default settings. Vocoder is basically a non-linear mixer.

Far as I got. I think this technique is trying to decode the show with the low tones as the lower sideband and the high tones as the upper. That’s why 800Hz as “carrier” seems to sorta work.

I tried Effect > Change Pitch to push it all higher, but that effect is so ratty that the error and distortion swamped any good it might have done. It sounded like gargling a sore throat.