Even out wonky/offset waveform (not normalise)

[unknownsailor]

Here is a single waveform of an entire project from Premiere Pro, consisting of multiple audio tracks:

I've highlighted a couple of areas where the waveform has bias above the 0db line (it occurs in both left and right channels but I've just highlighted the left channel).

I know there is a one-click solution to even out the waveform so that it looks more like the section between 3-10 minutes, but I don't remember what it is! Any pointers?

TIA.

[paulfigg]

I wrote this 9 years ago:

"Asymmetric Wavefirms: Should You Be Concerned?"

Make sure you recognize how DC Offset differs.

-paul.

@produceNewMedia

[unknownsailor]

Thanks for that, Paul. That goes some way to explain what is going on and now I know what nomenclature to use, which was half my problem with finding a solution. My issue is an asymmetric waveform.

The one-click solution I was looking for is under Effects/Stereo Imagery/Graphic Phase Shifter. If I set this to +80db it fixes some of the asymmetric issues, but not all.

I've gone back to each source clip and noticed that many of my audio clips are asymetric, which I hadn't noticed before. Here is a clip recorded on the GH5 with a Sennheiser MKE600 via the XLR adapter mounted on the hotshoe, gain and lowcut set to zero (I just increased db in Audition to visibly magnify the difference)

This doesn't look like a DC offset issue because, as I understand it, the bassline would be shifted above (or below) 0db.

So, what is causing this asymmetry? Is this just a manifestation of my camera set-up and to be expected? In which case, is it just a case of running Graphic Phase Shifter on every clip?

What is interesting is that in my original screen grab, the section between 3-10 minutes is also audio from the GH5, but that doesn't look asymmetric.

[paulfigg]

I wouldn't be overly concerned with the above example. As far as the root cause - 'could be any number of things including what I referenced in the documentation. Eg. it's common for a mixer supplied or mic supplied hi-pass filter to cause the anomoly.

You can certainly use the Audition supplied Graphic Phase Shifter tool to attempt to "correct" your clips. However it will require a significant amount of trial and error.

Best case is to use Adapative Phase Rotation in iZotope RX. And the processing option is also avaialble in Acoustica by Acon Digital. In essence there's no tweaking. Both tool options function as an offline adaptive process. Press a button and let it do it's thing. Note the least expensive version of RX includes the option. So there's no need to spend big $$ on the more advanced versions.

-paul.

[unknownsailor]

Great, I'll look into those, Paul.

FYI I quickly exported the offending GH5 sequences (which had already been edited) and applied Shifter/MultiBand Comp/Normalise, brought them back in as single wav files, then exported the whole episode, which you can see below:

Still not perfect but a marked improvement compared to the original image I posted.

In the future, I need to apply these fixes on the source clips immediately because in this instance I've ended up running the multiband compressor and normalise twice on the GH5 clips (effectively editing clips that were already edited).

Thank you for your help.

[paulfigg]

Cool and you're welcome.  I expect Steve to chime in and share additional insight.

-paul.

[SteveG_AudioMasters_]

I think that the only thing I'd add is about the causes of it - which from an acoustic point of view can be quite natural. You'll notice that in every single example that the signal has shifted in a positive direction. This is because a) there's universal agreement about the direction of air and electrical signals, with an outward flow being represented by a positive signal and b) there are very few sources of audio that produce an output from sucking (the harmonica being the only example that immediately springs to mind).

And that's what is behind this - the flow of air. You can't speak without breathing, and it's the forward flow that shifts the microphone's diaphragm. How much it shifts it depends upon a lot of factors, but basically it's to do with bass response, which is why a high pass filter can often make a significant difference. There are some causes that are often worse than the human voice; mainly these will be brass instruments like trumpets, especially when they are close-miced.

But - you can do quite a lot about this if it happens regularly by taking some simple steps. Taping a pencil vertially in front of a mic to act as a blast deflector will, as a secondary effect, reduce the dc offset. Even a normal pop filter will make a difference, especially with omni mics (these tend to have better bass responses, making them more susceptible to air movements).

And that's the thing really - it's always better to prevent the problem if you can, rather than just treat it.