I have several voice recordings sporting an unsymmetrical waveform (some parts already having the highest amplitude are heavily unsymmetrical).

The audio isn't very loud but can't be further amplified / normalized digitally as the samples in the unsymmetrical parts already touch the 16 bit range limit, despite no single part of the waveform actually use the 16 bit range by itself.

The unsymmetric parts are balanced in the mean, eg. they do not feature relevant DC offsets.

Is there any way using standard software tools to "rebalance" the waveform on those parts to get more headroom for amplification?

  • It might have to do with DC offset – user18803 Jul 21 '16 at 20:02
  • Most waveforms from natural sources (voices, instruments, etc.) are asymmetrical. But it sounds like you are talking about CLIPPING not ASYMMETRY. Those are two very different things. If you are saying that you are starting with audio that is already clipped, changing its level by amplification or attenuation won't fix the clipping problem. There are some software applications that can attempt to repair clipped audio waveforms with varying success. Please provide an image showing the waveform you are asking about so that we can confirm that we are all talking about the same thing and agree on – Richard Crowley Jul 22 '16 at 13:55
  • The waveform is NOT clipped right now, but any say 10ms part of does not use the full value range. If all peaks where symmetrical, I could boost this waveform by > 3db, however as it is asymetrical, I could not boost by any amount without introducing clipping. – dronus Jul 23 '16 at 17:04
  • No, there is no mean DC offset, the 'positive' parts of the signal have much larger peak values but are rare, the 'negative' parts are more frequent and have lower peak values. So the mean is almost 0. – dronus Jul 23 '16 at 17:08
  • It almost looks like f-video.s3.amazonaws.com/dc-offset-removed-large.png from producenewmedia.com/… . Only with the peaks just touching the limit. – dronus Jul 23 '16 at 17:09

I too agree that the question does not provide the right detail BUT talking about asymmetrical waveforms, the answer lies in phase rotation, the way kahn (or whatever spelling that name has) the radiowaves pioneer managed to squeeze +6 db headroom from male voices, cause this is very often in male vocal recordings (And class A amplifiers when they are driven hard)!

Another way of fighting this issue to gain some more headroom, is passing your signal through a hardware device with audio transformers!

First thing i would look for is the phase rotation plugin: play with the poles and bandwidth and in the meantime record your signal internally so you can see the results!

keep in mind, phase rotation can be found also as group delay(in audio applications), the term "phase rotator" is oftenly misused in plugin applications and what you get is plugins that just flip the phase or just ... phasers...

Here are some very usefull articles covering wave asymmetry and phase rotation as a counteraction:



Scientific: http://www.w3am.com/8poleapf.html

At last , you can also look for third party software that does the job, i bet you'll find some in form of plugin or not! :D

Good luck!

  • Nice answer, but unless you have to push the gain, I'd leave it alone. I have come across this before, and I needed the gain, so I just compressed it so it was more symmetrical. I can see how a phase rotator would work though. – Marc W Aug 3 '16 at 21:29
  • Well compression sadly is not a solution. It might give more gain because of compression and make up gain , but it is just a counter action. My opinion is tha you should first bring the waveform to a healthy state and then compress :) – frcake Aug 5 '16 at 20:13
  • Well it seemed to work at the time anyway! So do you think phase rotation could be used to combat things like phase cancellation when summing stereo to mono? – Marc W Aug 5 '16 at 21:03
  • Hmm i think what you describe is a different problem. Phase cancellation in mono "summing" should occur only if phase corellation between L & R is not correct ! Other than that when you have a healthy phase relationship you should hear the same thing only in mono. In stereo phase correlation sometimes is not that important and actually many times is desired as it results to stereo expansion effects. If you have a problem like that try posting a question with a sample , im sure it will be solved in no time :) – frcake Aug 5 '16 at 22:53
  • Also if you want to learn more about phase rotation i can provide the practical / creative details for you ! Its a very nice effect with various ways of use! – frcake Aug 5 '16 at 22:56

The unsymmetrical voice recordings you describe are perfectly 'healthy' and true to nature.

As already noted, a phase rotation tool should be able to move the high frequency, thin pulse-like peak back into the rest of the waveform.

Another method of doing this, is to use a frequency shifter, shifting by as little as a couple of Hz up or down. This is less effective, and doesn't sound great (transients are smeared)... but I've found phase rotation plugins difficult to find.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.