The loudness dilema

[~BURNY~]

I don't like squashed sounds. On the other hand, I don't want my mixes to sound
weak (espescially when it comes to samplers cd), plus a lot of musicians are naïve (they want their music hot ).
So I'm just trying to find the best compromise and it's a pain in the ass actually.
A squashed mix = Distortion generating harmonics resulting in harsh cymbals and mudiness and blurry feel + Dead transcients resulting in lake of punch and bad localisation/stereo spread + changed levels mostly resulting in nervous breakdown.
Here's my premastering routine (assuming my mix is fully satisfying):
32 bit float rendering.
Removing DC offset.
40Hz hi-pass filtering.
Gentle global EQ if necessary (80/300hz negative bel eq, High shelved @3khz or higher).
Fast compression gentle ratio.
DC offset.
Fast (multi-band) limiting until I feel distortion coming then back a bit.
DC offset.
Gentle global EQ if necessary.
I try to save 4/3db min dynamic.
Then bit depht conversion+dithering.

My mix are usually still softer than common commercial mixes in a way I feel I can deal with that (ok, musicians can't). It's ok And I'm already asking myself "where's my mix?". Just mud and harshness.

Is there something I'm missing? Do you guys have tips that can improve this a just a tiny bit?

 

[A Gruesome Discovery]

I don't see what the big downside is to having "the quietest song on the compilation" if it's also the best-sounding. I know there's a perception that Joe Listener will lose interest in a track if it's not 100% maxed out the whole time, but ultimately I think the goal should be maximum audio fidelity, and not trying to one-up the next guy with your louder mix/greener lawn/bigger car/Freudian analogies.
We've been granted 96dB of dynamic range with the 16-bit CD, and modern pop CDs are using one or two. I'm willing to bet that they'll still only use one or two dB when 24-bit is the standard (144dB of dynamic range, mostly wasted). We've all got volume knobs; if people want a super-squashed track, they can just turn it up to threshold of pain and let their ears act as brickwall limiters.
I know this doesn't answer your question, but I find the loudness race silly when applied to music that isn't meant for mass consumption. Master your music so it sounds the way you want it to, and don't worry about the other guys. You can't go above 0dB no matter what, so the only way to compete with something that has 1dB of dynamic range is to slam your mix in a similar fashion.
Having said that, if you still want to sqeeze some extra volume out, make sure there's no particular frequency band that's stealing too much dynamic range, especially in the lows; a carefully set multiband compressor should take care of any offending frequencies without altering the tone as much as a regular EQ cut would. But you should first decide if this is really what you want to do; if you're happy with the overall tonality, then don't bother squashing; or perhaps you could just use a multiband comp on a special "compilation only" mix.

 

[~BURNY~]

I totally agree with what you said, just trying to find the best compromise to satisfy the musicians without killing my work. The compilation argument is a bit childish, I agree.

 

[A Gruesome Discovery]

Oh, also, what are you using for a limiter? I've used a few, and have found varying degrees of quality among them. Some just seem to get gritty when pushed.

 

[Brett - K A L I S I A]

Sorry if I get it wrong, but you remove DC Offset and apply EQ after the limiter ???

 

[~BURNY~]

I remove DC offset as long as the software find some at any stage. And it usually find some.
The final eq (if necessary) is intented to compensate possible harshness and is always negative.
The limiter I use is often Waves L3 (I just found it "not so bad" but not perfect). I also like my old single band timeworks mastering compressor (don't know why they didn't call it limiter...).
I think I will buy a finalizer express as soon as I can afford it. I'm planning to borrow one from a friend to try before I make my final decision.

 

[Seizure.]

what the hell is DC offset?

 

[~BURNY~]

DC offset is when your waveform is not centered on the 0 resulting in possible dynamic loss, distortion and calculation error (correct me if I'm wrong)

 

[Brett - K A L I S I A]

But even negative eq can result in some frequencies being boosted and thus possibly going over 0db, so using it after the limiter is a no-no, same goes for DC offset, limiter MUST be the very last process !

 

[~BURNY~]

Ok, I knew about frequencies being boosted (excursions de filtres, don't know how to say it in english) but I was adjusting the output control accordingly. Now I realize I should not do this and just get the cleanest signal possible. As for the DC offset, could you be more specific? As long as the signal doesn't reach the 0dbfs, I can't see where it hurts the signal... And don't you think it could affect the bit depth conversion ( or maybe the burning process)in some way if there's still an offset? Just wondering... Anyway thanks for the hints. Much appreciated.

 

[Brett - K A L I S I A]

Ok, I knew about frequencies being boosted (excursions de filtres, don't know how to say it in english) but I was adjusting the output control accordingly.

Then I guess you're maybe losing a few "precious" dbs (or maybe just tenth of db). Anyway, since you use L3, you can use the "built in EQ" to compensate for this harshness, that would be my advice, or put the EQ before the L3 of course. By the way, quick question here for you and everybody else : where do you take your limit ? I usually go to -0.3db because I read that some cd player can clip over this, but is this true ?

As long as the signal doesn't reach the 0dbfs, I can't see where it hurts the signal... And don't you think it could affect the bit depth conversion ( or maybe the burning process)in some way if there's still an offset? Just wondering... Anyway thanks for the hints. Much appreciated.

Let's say you have your peaks at 0db and you have a DC Offset. Compensating for DC Offset is like sliding your waveform (up or down - for this example I'll say up by 0.1db), so some peaks in the upper waveform will be increased, thus overloading. Of course, there are very few chances that this can happen (especially since usually DC Offset is very very small), but I just thought this was a bit weird to do (and sounds like a pain in the arse)... Aren't you a bit anal about the DC Offset thing ? lol I'd say "Only remove it once" before the mastering stage (unless you have very big issues with it when you track). Take some ASPA (Andy Sneap Produced Album), rip it in Wav on your computer, and check out the DC Offset... It's really not 0... But still...

Anyway, I guess this is not a big problem AT ALL, I just wanted to warn you that it seemed weird to me, but if it works, that's the most important.

 

[~BURNY~]

Aren't you a bit anal about the DC Offset thing ? lol

I guess you're right...