Audacity Forum

I am trying to take the harshness out of an mp3 file that was apparently recorded at saturation, producing an extremely clipped waveform for virtually the whole recording. The only things that were somewhat successful were: 1) Noise Removal at the lowest removal setting, though that produced some flanging, and, 2) Running successively both the High Pass filter and the Low Pass filter at about 3000 kHz, but still left some clipping effect.

What I am curious about is this - inside the dark blue clipped waveform is a lighter blue waveform that appears not to be clipped, as though that would be the waveform contour if the recording had not been clipped. Is there any way to get Audacity to recover the unclipped lighter blue waveform inside the dark blue clipped waveform? I have searched the forum posts and Audacity's Help file for this and come up empty. Appreciate any help you can give me!

Clipping is usually one of the fatal problems. Even if you manage to "rebuild" the waveforms, the performance will change, but not always sound any better.

So you had analog clipping, right? If you see waveforms going over the clipped signals, then an audio board, processor, or transmission system failed before the show became ones and zeros. That can actually be worse, because analog audio systems frequently clip "dirty" instead of straight across. You may be watching some of the dirt.

The only show I was ever able to rescue was a stereo show where the two halves of the conversation were on different microphones. The thing was so overloaded, I was able to get her voice just fine on his microphone.

Koz

Someone correct me if I'm wrong (I'm looking at you, Richard).

The dark blue part of the waveform shows you the highest peak in the area that pixel represents (remember that zooming out far enough forces Audacity to display many samples using only one column of pixels, so each pixel might represent several hundred samples, hence they show you the highest one).

Each light blue part column displays the average RMS value for that same set of samples. That's not useful as audio data, it's only useful as a rough gauge of how loud something might sound.

In other words, you're recording is fubar'd.

In video we have the bad joke of applying the "reshoot filter."

Koz

"Reshoot filter", huh? LOL!! Wish there was one I could download! Thankfully it wasn't my recording, just a talk recorded on a web site I frequent. Would have been nice to hear it un-fuzzed, but you can't win 'em all. Thanks for the info!