Don’t know if it’s already been known here but I just found out chopped top waveforms by compressors/limiters for loudness on commercial CD music can regain their shape after first reducing amplification and then applying EQ edits in Audacity.
Below is an Audacity screengrab of a zoomed in chopped top waveform on the Electric Light Orchestra’s “Do Ya” song off their 1976 “New World Record” album remastered in 2006 by Sony engineer Joe Palmaccio. The waveform below shows the repaired results after first reducing amplification by -4db and applying an eq using Apple’s AU GraphicEQ to boost kick drum hits by 4db (40-80Hz) region and boosting highs by 2db (2K-6KHz). I reduced amplification by 2db afterward and applied 4db increase limiter using George Yohngs W1 Limiter.
I take from this that digital audio wave files are quite pliable even after what appears damaging edits. I can say I can’t hear any artifacts in the results after doing this.
Also it must be noted that just reducing amplification and then applying the limiter does not fix anything. Both A/B waveforms show the chopped top appearance though I do have to say that limiters aren’t as damaging as I’ld thought just by visual examination of the waveforms at this magnification. I also didn’t apply the EQ without first reducing amplification to check headroom volume on Audacity’s 32bit floating point processing.
Am I missing something in arriving at this new understanding on the resilience of editing .aiff audio waveform files?
Just applied the EQ on the unedited original file and I’ll make it official, Audacity’s 32bit floating point has at least 6db of headroom which was how much I had to reduce amplification to take away the red wall of clipping.
Waveforms look identical to the bottom screengrab version. No chopped top waveforms.
Sorry, but you are NOT “repairing” clipping. Audacity has a Clip Fix effect, which attempts to repair clipping.
Information is permanently lost when the waveform is clipped (it’s impossible to know the height or shape of the part that was clipped off). But in some cases, you may be able to make a good approximation. When I tried it on a commercial recording, the waveform looked better but it didn’t fix the distortion.
You’re seeing an altered wave shape which is a result of the filtering (EQ) and phase-shifting different frequencies by different amounts (a side-effect of filtering). There is also something called an all-pass filter that shifts the phase of different frequencies relative to each other without affecting the sound. You get a little filtering and quite a bit of al-pass filtering if you make a vinyl record from a clipped file, leading many people to believe that the vinyl copy is has less (dynamic) compression than the digital version (even though they are sometimes made from the same master). Or if you make an MP3, you’ll get similar results with slightly boosted peaks (actually some peaks are boosted and others reduced).
I can say I can’t hear any artifacts in the results after doing this.
Are you saying you can’t hear the EQ?
Also it must be noted that just reducing amplification and then applying the limiter does not fix anything.
There are limiters that will do that but you’d have to decide if the additional limiting/compression is worth it. The “problem” with the Audacity limiter seems to be that you can’t change the release time (hold time) to zero… You’d like to “instantly” alter the wave shape moment-to-moment.
And actually, you shouldn’t have to reduce the amplification first. It depends where you set the limiter’s threshold/limit. Then if you do push-down the peaks below 0dB, you might want to re-normalize (i.e. Apply make-up gain).
Audacity’s 32bit floating point has at least 6db of headroom
For all practical purposes, 32-bit floating-point has unlimited headroom.
Your comment that I have NOT repaired clipping is a bit confusing since I didn’t say that was my main purpose. But then you may be referring to this comment I made…
The waveform below shows the > repaired results > after first reducing amplification by -4db and applying an eq using Apple’s AU GraphicEQ to boost kick drum hits by 4db (40-80Hz) region and boosting highs by 2db (2K-6KHz). I reduced amplification by 2db afterward and applied 4db increase limiter using George Yohngs W1 Limiter.
Technically according to Audacity clipping indicators there was no clipping in the original commercial aiff file off the ELO CD, just the chopped top waveforms none of us will ever know what was “clipped” from the original waveform before the audio engineer had at it. That waveform data is permanently gone. That’s not the clipping I’m talking about. I’m referring only to how chopped top waveforms can be reshaped through EQ’ing without distorting the original sound. With the EQ I’ve reshaped the sound characteristics of the flat but overly midrange heavy Palmaccio remasterings by putting back the beef in the kick bass without distortions. The sound is much improved especially playing in my car with two subs.
There is the original Gastwirt CD remastering versions that don’t have the loudness war edits applied which I don’t have of this piece, but I’m going to examine the two remastering jobs on those where I do have both to see how much data was lost by the chopped top remastering versions.
Audacity’s 32 bit floating point processing may have endless headroom but my Mac OS system volume control slider may disagree. I’ve applied extreme EQ edits to the original loudness war versions without first reducing amplification in Audacity and I do get distortion going too far in the EQ editor. But then I’m not trying to make it louder but just change or improve the sound characteristic of these ELO songs.
I’ve tried out Audacity’s built in limiter. I really couldn’t hear any difference compared to Yohng’s W1 which is a lot less complicated. Audacity’s limiter has some interesting options if only I could understand how to work them and tell what they do to the sound on edited .aiff commercial CD music. I just want to set it and forget it. Both limiters do seem to have trouble keeping low bass frequencies around 50Hz and below from getting out of hand.
I can hear a couple of low end booms now and then on my car’s subs in the trunk but not my headphones or my home stereo speaker system. That’s when I have to back off the EQ, but I can’t keep up with every errant low frequency bottom end rumble that raises its ugly head but at least I’m learning how to check for it before I burn the edited songs to CD. If I see a -3 to 0db in a spectrum check on these errant bottom enders after setting the limiter threshold to -4 or -6db I know I’ll have to edit them down individually. They don’t harm my main speakers since they’re on a 12db per octave 80Hz crossover network.
Also I just checked the -18db RMS Gastwirt remaster waveform against Palmaccio’s louder chopped top -12db RMS remaster and there’s not much wave peak shape difference lost. And on top of that there’s very few of these chopped tops throughout the waveform of these songs that can be heard unless one can hear it within the sound duration of a snap.
A multi-band compressor, set to compress the bass, would catch all of them.
I don’t speak Mac, so I don’t where you can get a multi-band compressor plugin.
However, within Audacity, you could use the De-Esser plugin, to De-Thump,
(although it wasn’t designed for the purpose), if you set it to operate in the bass range.
Thanks for the multi-band compressor tip. I’ve used the Apple supplied AU MulitbandComp in Audacity but it’s not as precise and specific for controlling bass. And it wasn’t as good as I thought applying it as an overall EQ tool which with its simple frequency band db settings made it easier but less precise which tended to give the music a kind of overly bright and crispy fake sound over using a straight EQ.
And besides that it’s pretty hard to hear and locate these low end thumps with headphones. I’ld have to scour the entire song with a spectrum analyzer to locate only the few and short duration thumps.
