mastering tip for us home types

[giraffe]

i read this somewhere a while ago, but i just tried it yesterday.

when using something like an L1, chain the bastards.
two of them at -3 sounds about a billion times then one at -6
(or would that be +6?, whatever)

but seriously, do that........ it's sooooooooooo much better.

two in a row, both doing less work.

really, do it now.















seriously.

 

[giraffe]

oops, i thought i was in the mastering foroum

 

[mshilarious]

oops, i thought i was in the mastering foroum

That's OK. I like compression then limiting, seems even more gentle, although 6dB is a buttload no matter how you slice it.

 

[Ford Van]

I am always a sucker for these kinds of test. LOL

Don't look at the files in an editor. Just play them tell me which is which.

One of the just used one L1 doing 6dB of gain reduction.

The other used two L1's, both doing 3dB of gain reduction each.

Which was which?

http://www.phoenixlightandsound.com/Audio/a.wav
http://www.phoenixlightandsound.com/Audio/b.wav

I actually had done this experiment many years ago.

 

[bigwillz24]

my guess a = -6dB and b is 2 L1's @ -3dB

 

[Dogman]

I think I like A better, but don't know the answer. A seems to have more punch, but it's close.

 

[giraffe]

i think b sounds better..........
there seems to be less strangle on the snare after the gits come in, and the hat sounds better to me too...


holds breath



edit/disclaimer: and that's on my friends logitech comp speakers

 

[Ford Van]

a.mp3 is two L1's.

b.mp3 is the single L1.

The peak rms is different between them, which will probably account for the difference.

 

[Ford Van]

So, I guess maybe two L1's really doesn't sound any better doing 6dB of gain reduction?

 

[giraffe]

ohhhhhhhhhhhh


shot down.
gota hate that.

what's your opinion on the whole thing.

 

[giraffe]

i'll post the song tommorow that i tried it on, both ways.
try one more time, but maybe it was all in my head.

 

[Ford Van]

I never really noticed much difference. Slightly at REALLY HIGH gain reductions, but, if I have to gain reduce that much to get the rms I need, the mix is WAY out of whack.

For up to around 6dB of gain reduction, I think just running one instance is fine.

 

[giraffe]

at the risk of beating a dead horse

here are the 2 samples i promised.
i don't have a place off hand to host .wav, so i hope 320kbps is good enough.

sample m
http://www.lightningmp3.com/live/file.php?fid=2612

sample k
http://www.lightningmp3.com/live/file.php?fid=2613

the entire song, lower bit rate (for any one who's interested)

song is "what i see"
http://www.lightningmp3.com/live/file.php?fid=2610

 

[mshilarious]

here are the 2 samples i promised.
i don't have a place off hand to host .wav, so i hope 320kbps is good enough.

sample m
http://www.lightningmp3.com/live/file.php?fid=2612

sample k
http://www.lightningmp3.com/live/file.php?fid=2613

the entire song, lower bit rate (for any one who's interested)

song is "what i see"
http://www.lightningmp3.com/live/file.php?fid=2610

M sounded a little nicer, I just liked the snare and hat a little better. Not a huge difference though.

Then I took K and added +24dB with the UAD Precision Limiter. Got to -8dBRMS, and as a bonus, every hat hit was a different volume! But I figured hey, why not louder? So I switched the release to 0.01msec., and added another +24dB: -6dBRMS, that's more like it

Sounds like total crap, but I learned one interesting thing: it's an easy way to get "When the Levee Breaks" sounding drums. Who knew?

 

[Ford Van]

I gotta tell you the truth man, I threw them both into a montage in Wavelab, which allowed me to have them playing at the same time, and I can solo each (when you solo one, then solo the other, the first solo'ed track doesn't stay solo'ed).

I didn't hear even a little "blip" when I would switch between them. I even went as far as to assure that they were sample accurate in playback, and had them both play at the same time. Predictably, I had to lower the master fader EXACTLY -6dB to keep from clipping. There was NO phase problems or anything when I did this. It just sounded like the same track with both playing at the same time.

I don't think there is ANY advantage is running the L1 twice or running it once to achieve the same RMS.

Sorry. You are just hearing a difference because you think there SHOULD be one. Now, there IS a slightly difference between the sample levels of the files, which is obvious when you zoom in close, but again, it is SO little of a change that I just don't think your ears could really make that out as a difference.

 

[mshilarious]

I gotta tell you the truth man, I threw them both into a montage in Wavelab, which allowed me to have them playing at the same time, and I can solo each (when you solo one, then solo the other, the first solo'ed track doesn't stay solo'ed).

I didn't hear even a little "blip" when I would switch between them. I even went as far as to assure that they were sample accurate in playback, and had them both play at the same time. Predictably, I had to lower the master fader EXACTLY -6dB to keep from clipping. There was NO phase problems or anything when I did this. It just sounded like the same track with both playing at the same time.

Clearly there should be exactly a +6dB peak on mixing, because all the peaks are occuring at the same places. However I expect the release behavior to be different between the two limiter setups, and that could account for the difference.

I lined up the tracks and flipped the phase on one; they don't null although it certainly gets pretty quiet

 

[peritus]

it's an easy way to get "When the Levee Breaks" sounding drums. Who knew?

Though you most likely know this, the drums on that song were recorded in a stairwell.... Interesting thread ya got here!

 

[giraffe]

totally possible it's all in my head

m is 2 L1's, and ever so slightly lower rms, maybe about .5db or something.
i still say m sounded better, but it was more evident in protools, the hi's in the git seemed a little strident in the single L1.
after a dither and data compression i'm not sure if i could tell them apart any more if you mixed them up.

which begs the question, how little of an improvement is too little to bother with?

or is that the vaunted "stupid question"?

 

[Ford Van]

totally possible it's all in my head

m is 2 L1's, and ever so slightly lower rms, maybe about .5db or something.
i still say m sounded better, but it was more evident in protools, the hi's in the git seemed a little strident in the single L1.
after a dither and data compression i'm not sure if i could tell them apart any more if you mixed them up.

which begs the question, how little of an improvement is too little to bother with?

or is that the vaunted "stupid question"?

Okay, in ProStoolz, you are indeed going to start getting some weird stuff. While many other applications work 32 bit float (or better) throughout the application, ProStoolz uses 24 bit fixes for it's "insert" on the mixer.

Going between a 32 bit float and 24 bit fixed data path a couple of times can do some weird things.

 

[ryanlikestorock]

Two things should change when using two passes instead of one.

1. The program will be used twice, which makes the signal chain that little bit longer and theoretically worse.

2. The ratio of the compression/limiting will change exponentially - but, given that the L1 probably works very close to a ratio of infinity:1, I don't imagine you'd hear the infinity x infinity, seeing as it should also equal infinity.

In mastering, I usually keep the levels of the rest of the chain somewhere near 0db so the limiter doesn't have to do much work for gain. Some pieces of gear (or software, even) sound better for gain than others.