Asymmetric Gain makes a vocal louder and harsh (a little distorted)

[des0free]

Since I can now process audio tracks computationally, I can try weird math-calculation processing effects.

Looking at one of my vocal waveforms I noticed it was asymmetric. The normalized max reached +1 at one place, but the min was only "-0.5" (whereas, pure sine wave tone would vary between +1 and -1). So I wondered what if I multiply only all the negative points in the signal by 2?  Then the maximum signal would be +1 and minimum -1, so it would become more symmetric.  I found this makes the vocal louder, but also makes it harsh.  The signal is not clipping  but distortion has been introduced.

I doubt this would find much real use, unless one wants to make something sound both louder and HARSHER (distorted in a rather nasty sounding way).  Maybe there could be some rare use for a shy person with an ultra-smooth voice who is trying to sing in a nastier voice for heavy metal?   

First sound file/plot is before processing, second file/plot is after processing:

[StephenM]

More cool stuff Doug....I dig your wanting to...your drive toward experimentation...I kind of would like to hear a sample ...you'd be surprised what we would use even if but short bursts...
Wave graph sure does show it... I wonder if you tweak the math a bit...maybe some crazy function formula that could be adjusted? 
Math is fun and at one time I was quite good at trig, alg, geometry,  HS stuff...has tons of uses in industry...I never dug much deeper though.

[AndyR]

WOAH!!!

I think you might have unlocked something for me.

DC Offset. I believe that's what this asymmetry on a waveform is called?

My DAW (and Audacity, I believe) has the ability to remove DC offest from audio.

How it does it, I've no idea, I'm guessing not asymmetric gain if that adds distortion?

But anyway - if your audio has an offset, it kind of restricts how high it can be amplified before clipping occurs... In other words, it restricts how loud you can master without excessive compression/limiting!!! (which is the demon I am fighting with at the moment).

I shall be experimenting with the DC Offset thing later

[AndyR]

Awwww ... It's not real DC Offset ... it was me seeing 2 + 2 and thinking I can have 5! LOL

Some documentation (including Audacity's) can lead you to believe that asymmetric waveforms are suffering from DC Offset. But it's not the documentation's fault per se, just the level of knowledge (or not) the reader has....

Also PreSonus's Loudness detection on it's Project (mastering) page reports DC Offset. There's loads of users on forums going "omigod should I be worrying?"
(I hadn't even cared about the numbers I can see there - I probably will now!)

Asymmetric waveforms happen (eg human voice) it's fine, nothing to worry about (I found a nice SOS https://www.soundonsound.com/forum/viewtopic.php?f=16&t=74941 thread on it)

Actual DC Offset is a thing, and happens with faulty or mismatched hardware in the signal chain - it's a voltage. This will cause problems in mastering - low level distortion, clicks between songs with different offset values.

But this asymmetric waveform, which I see fair bit myself, is not an issue.

In PreSonus, I used the "Block DC Offset" in the Mixtool on the stereo mix and exported it. Waveform looks a little different and the loudness is slightly less. Re amplifying it to 0 produces a very similar sound, slightly quieter than the original, but otherwise, no change or ability to increase the over all volume. I'll need to experiment more to see if I can make use of it in the mastering chain to push louder without getting the bad vibes I'm getting at the moment.

In Audacity, you can remove DC Offset with the Normalise tool (switch off the Normalise option). I tested on the same audio and... Audacity does nothing at all! (I think I prefer this to what PreSonus's Mixtool does).

EDIT: From reading further on the PreSonus forums, where the users are going "is it a problem? what is it? etc" ... and no-one is really answering them... I suspect now that you can squeeze a tiny tiny bit more volume out by removing the reported "dc offest", but you need to do this by better use of eq choices and high pass filters earlier on (when mixing). What the PreSonus thing is actually reporting is a measure of the "asymmetricness" of the audio ... and if you want to go down the rabbit hole to try and force it to -infinity you can... but actually I suspect that filtering and limiting during the mastering process is doing the same job...


EDIT 2: I was about to ask for an example of your asymmetric gain... then I spotted the attachment on the op
Ah, I see what you mean. Nope, not quite for me neither! LOL
(or at least not on its own)
From what I've read now, I think that the source is meant to asymmetric, and the asymmetric gain is possibly kind of "unnatural".
It sounds a bit like what I get with "excessive limiting" but in this case it hasn't lost the top end.
Actually, it sounds a fair bit like some of the "parallel compression" tracks I use to augment a lead vocal - copy the lead, compress it to death and add highs. When you put this in the mix at a low level beside the actual vocal, you might like how it makes the vocal more "present" in the mix, but you can actually turn it down and hear more of everything else (and this might widen the stereo of the mix too - lowering the mono vocal can make the left and right more distinct).

[Geir]

In theory, if you have a low-cut (high pass) filter on your input you should avoid this problem all together I think

[AndyR]

Hi Geir!

Yep, that's one of the things I found in my reading...

My current song is PLASTERED with high pass filters - in fact I've been using them for years on the BR too.

Actually, there isn't one on the Main stereo out channel...

But I've just looked in the Loudness Information in the mastering Project that works on the results of that stereo out, and it looks good.

Both Left and Right channels are the apparently "coveted" -infinity for DC both before and after mastering effects.

The waveforms still look "asymmetric" in Audacity though!

I'm going to stop worrying or even thinking about it! 

My current issue is that my first completed recording from Studio One sounds utterly utterly fabby but, even after my mastering, is LOADS quieter than the last two I did on the BR1600.

And then the last mastering attempt on Saturday seriously compromised all the fabbyness of the mix.

I've seen/heard enough to know that I CAN do it with the recordings I've got in the DAW, I've just got to learn how...

So I'm going back to learning more about mixing/mastering and trying to discover what I was doing on the BR1600 that I've missed in the DAW

Also I've found some superb exclusive content on the PreSonus Sphere website. The ones I'm watching at the moment are from a Mastering Engineer, explaining how to get a mix ready for mastering. I've pretty much failed all his points and I'm less than halfway through it... So he's kind of saying it's all in the mix not being ready for mastering, which I suspect I'm going to have to agree with in a few days!

Anyways, always hoping for a shortcut, I thought reading des0free's original post was giving me the secret of loud mastering... alas no!!