Identifying high-frequency interference

I’ve gotten fairly good with Audacity but have not tracked down a relevant tutorial for this new problem. I have a voice recording made in 1964. The person used a real-to-reel but lacked training. The most obvious problem is sporadic high-frequency squeal. The recording was transferred to cassette then CD, and I extracted WAV files from that. I easily got rid of some rumble. Then I used something called a Mag Notch Filter to zap the squeal at about 1,600 Hz. But there are some overtones lurking, and the only way I know how to identify their frequencies is to play whack-a-mole with the EQ. I tried the plot spectrum tool but don’t feel confident. Tips?

Plot Spectrum can be used for this and you can change the view with the “size” control to get something useful. This is a 2000 Hz tone viewed three different ways: With a low sample size which give a very sloppy display, with a medium sample size and then with a very, very high sample size. You can just see the 2000 blue spike in the picture.

Obviously, the high sample is the most accurate, but it’s almost impossible to see, particularly if you have a lot of tones. You should know that Analyze Spectrum gets more and more accurate as you drag the window bigger.

Click each forum picture to make it bigger.

Screen shot 2013-06-20 at 3.02.48 PM.png
Screen shot 2013-06-20 at 3.02.18 PM.png
Screen shot 2013-06-20 at 3.02.01 PM.png

Thanks for the explanation and excellent screen shots. Seems like I’m on the right track!

OK, now I’m going to dump cold water on it. If you have a distorted tone, a buzz or mosquito sound, that’s going to be a lot more interesting. Tones that are not pure tuning fork or flute sounds get their richness and complexity by having more than one spectrum spike. Sometimes a lot more. The good news they always go up, so if the interference is already a very high pitched whine, you may be able to get rid of almost all of it without destroying the show.

This is a single piano note. I think it was three octaves down from the middle of the piano.

The big peak is at “G,” but the rest of the spikes are the overtones and harmonics. They are the reason you would never mistake that note for anything other than a piano.


Actually this is working quite well so far. The “squeal” is obviously not part of the frequency range of the two adults talking to each other. Just learning the finer points of making a notch filter do what I want (and no more)! The family member I’m doing this for thinks it’s a miracle that an obnoxious noise can reasonably be removed like this. Audacity is making me look like good.