The dubstep tips super thread

threnody · Post by **threnody** » Thu Apr 06, 2006 11:53 pm

This is a really interesting thread. lots more to say later but....

as for this normalizing. The reason mastering plants say do not normalize is it makes their job very difficult. They put compressors, EQs, limiters and maybe some spectral emphasis onto a finished track. If the track is already normalised this is hard (especially EQing) as they cannot boost frequencies which are already at their peak. They could simply turn the track down but the noise floor rises at the same amount as the waves (so to speak). Their corrective procedures can bring up the good sounds whilst reducing or leaving the bad (noise) ones.

Normalizing should always be the last step taken to make the track as loud as possible so if you have finished a track you are sending to labels or djs without mastering before then normalise. If you are sending it to be mastered then it is not the last stage for that track.

It is always worth normalizing your samples as you then have a good solid base for your mixdown and can turn stuff down rather than up.

+ the reference to limiter, compresser, eq is meant about the finished track as an after process not to individual sounds in the track.

doomstep · Post by **doomstep** » Fri Apr 07, 2006 4:10 am

KION wrote:It aint anything to do with what 'the kids' are for or against. Its basic "textbook" knowledge man.

. . . well these things are as fashion driven as anything else at the the end of the day, the science stays the same, so t speak, but peoples perceptions / techniques change as tech. n listening adapts. I was just curious about this as I hadnt come across it . . . alot of my 'textbook' knowledge is 5 years old now. The widespread availability of 24bit DAWs and cheap CDR tech has really changed things in recent years . . .

I dunno if we're all using the word normalize in exactlly the same fashion, I'd be interested t know how normalization in itself is inherantly degrading tho . . . anymore degrading than changing the volume of a file.

threnody wrote:The reason mastering plants say do not normalize is it makes their job very difficult.

Very logical . . . when u think about it huh. I assume this means dont normalize to 0dbfs ? as that would really make things hard for the mastering eng. Fat guy in a little coat style

basically adhereing to the logic in this point :

* Try to keep the peak levels of your mix under -3db. This is very important.

. . . at the end of the day tho emalkay is bang on, but I reckon the L1 has def. given mastering eng.s a few grey hairs in recent times.

So is anyone working wif analog compressors like me or is it all plugs ?

kion · Post by **kion** » Sat Apr 08, 2006 11:47 am

Regarding fashion, normalizing a mix has never been good practice. But I know what you mean about limiting/compression, as that is a fashion thing.

For the record I wasn't talking about normalizing on individual samples - a lot of people do that and that seems to be acceptable practice (some limit individual samples instead of normalizing).

But I don't normalize a finished mix. If I'm sending out tracks I do master them myself after bouncedown (in Wavelab) using a combination of linear multiband compression, linear EQ and multiband limiting. Just to get a decent level more than anything.

Normalizing an overall mix doesn't actually do anything sonically positive regardless, as it just increases the transient peak levels, not the RMS, and psychologically doesn't sound 'louder' to the human ear (unless you use the RMS setting, which is not much good either as your mix will usually clip!)

For your own use when playing out (if not cutting but just playing from CDr), you're better off using a mastering limiter offline.

But, If you do want to run a limiter on your master buss - especially if throughout the mixing process, its gonna distort your mix-balance perception. Best to get the mix with few transient spikes first, then run the limiter if thats the way you wanna bounce it.

A good way of balancing the sounds of your mix is as follows:

After the mixdown process is complete (or during it if you want to check the balance) bounce your track without the limiter first, check both sonically and visually for any spikes by importing your bounced file back into your daw (and line it up with your sequenced track), mute the wav/aiff and play the track.

Find the spikes in your wav/aiff and play the track from those positions so you can see/hear what instruments/samples are causing the spikes, and either compress those audio/instrument channels a little more (to keep the sonic level but kill the transient), or just turn those channels down a little.

Rebounce and repeat the procedure until you've got a well balanced mix.

When satisified, if limiting your mix in your daw, turn the limiter on (on the master bus), and bouce your track.

drewdrops · Post by **drewdrops** » Mon Apr 10, 2006 3:28 pm

great thread, mastering info was v interesting....

paradigm_x · Post by **paradigm_x** » Mon Apr 10, 2006 4:33 pm

totally with kion here

digital systems are more accurate at the top end of the loudness scale.

ie if you have a really quiet sound and normalize it, you dont gain any resolution, you get a 'blocky' but louder sound.

so if youre mix is quiet, dont normalize it sort it out in the mix, change the gains in cubase whatever, til it peaks at just under 0 dB.

i also think putting a limiter on the main out is a bad idea it makes you lazy in terms of balancing because it never clips. Ive got shitloads better at mixing recently and with a bit of work you can get an almost rectangular (ie limited-esque!) mix just by finding the perfect space for each sound. I also think that mixing and mastering should be attempted seperatley. Get your mix as good as possible, then get the peak to just under 0 dB. bounce it out, then leave overnight, then have a go at mastering the next day. If there are problems that require loads of work at this stage, go back to the mix and fix it. I find just a dB or two of limiting is all you need, with a little bit (1 or 2 dB) of overall eq sometimes.

my 2c

peas

bze · Post by **bze** » Tue Apr 11, 2006 8:56 am

Well if any effect is applied to the final mixdown it should be definately done offline. No way I'm going to put anything on the master channel of my tracks. Only a very soft limiter and that in extreme testing purposes only.

Interesting read tho, I might have even learned something from this. Have to do some experiments when I actually have something finished.

paradigm_x · Post by **paradigm_x** » Tue Apr 11, 2006 10:27 am

yeah sometimes it can be ok to put a limiter on the master when youre playing with the tune, just to get an idea of what the limited 'mastered' version would sound like. But when you get to the serious mixdown stage, get rid of it. IMO etc...

cardio · Post by **cardio** » Wed Apr 12, 2006 5:20 am

if i normalize while rendering (out of ableton live in my case) doesn't it just boost the master volume as much as it can with clipping?

i can't think of how there could be anything wrong with this--- anyone?

subframe · Post by **subframe** » Wed Apr 12, 2006 5:43 am

Here's the deal. If your audio is in integer, you will get aliasing when you normalize. Each sample is described by a number from a range of possible values (that range depends on your bit depth). Your software you use to write tunes will either use integer and thus always be accurate, or work in floating point, in which case it doesn't matter (see below) However, when you normalize, you simply multiply the value of every single sample by some percentage. For many/most of your samples, this will result in a non-integer number, which must be rounded to the nearest integer. This rounding introduces errors i.e. aliasing.

While the same issue exists with floating-point processing as well, the aliasing is likely to be negligible; it is likely to be less than a DAC will be capable of picking up.

Thus, normalization is bad.

[edit] here's some cool words on aliasing, btw: http://en.wikipedia.org/wiki/Aliasing. The Waldorf XT actually allows user control over how much aliasing you want in your sound. Have I said how much I love my XT? [/edit]

doomstep · Post by **doomstep** » Wed Apr 12, 2006 9:50 am

wikid, thanx heaps for that subframe . . . so really tho, if you're normalizing in floating point then normalization is only theoretically bad, but could still be usefull . . . insomuchas we listen t music through DACs, not inside our computers, and the increase in volume, for artistic, or technical reasons could be whats needed . . .

I think it helps t remember that as well as mixing 24(or woteva) trax down t 2, you're also mixing 24bits down t 16 . . .

Paradigm X wrote:ie if you have a really quiet sound and normalize it, you dont gain any resolution, you get a 'blocky' but louder sound.

. . . this depends on the bit depth of what yr normalizing & the intended bit depth of the final product, it also depends on how the sound was genetrated or if it was recorded, through what n with what . . .

but yeah if you've got a really quiet 16bit file then normalizing is jus gonna give you a loud lo-res sound, cos really its not a 16bit sound yeah

paradigm_x · Post by **paradigm_x** » Wed Apr 12, 2006 11:11 am

well, no, the maths are the same no matter what youre doing

but i agree, subjectively most of the time no-one would ever notice. in a club at 3 in the morning i cant see many people going home because of aliasing errors...

i just think its a good habit to get into, do things at the highest level possible (digitally) and dont normalise.

lets agree and finish it there !

peas

doomstep · Post by **doomstep** » Wed Apr 12, 2006 11:18 am

oh sorry, i thought it was a disscussion . . . not some sort of competition . . . my mistake.

paradigm_x · Post by **paradigm_x** » Wed Apr 12, 2006 11:23 am

oh and back on topic so i contribute;

stick something in before the delay on your fx send - eg a pitch shifter or glitch or something, keep the original clean and get some messed up fx on it. blend to taste

vintage warmer sounds mint on bass

keep your subs delay and fx free, it makes it muddy. Phase difference will mean all kinds of unpredictable adding and subtracting in the wave.

a good tip is when mixing, if you solo just the drums, and then just the bass they should be at approx the same level

cut the low (and high) out of any sounds that dont need it. Eg a string part, youd be surprised how much crap is floating around at low freqs, hipass it to save room in your mix, and avoid interference with bass.

Cuts on EQ are always better than boosts, remove the shit, dont boost the good bits (save space again)

I go on about space a lot, there was a really good tutorial on eq on dnbscene, which has sadly died, im trying to get a copy again. It makes you visualise a mix as a box inot which you can fit various elements. It was very good, im hoping to find it and rehost. BAsically if you think of a box with a lid, and the depth eqauls the loudness and the width represents the frequency. If you have say a huge bass and a huge kick, youll 'poke thru the top of the lid'. Its hard to explain well but this tutorial was great.

NB nothing is gospel, break any or all rules, thats kind of the point, but these are some things ive found helpful.

cheers

paradigm_x · Post by **paradigm_x** » Wed Apr 12, 2006 11:24 am

doomstep wrote:oh sorry, i thought it was a disscussion . . . not some sort of competition . . . my mistake.

sorry ?

so did i

im sorry if i offended you didnt mean to.

peace

kion · Post by **kion** » Wed Apr 12, 2006 11:29 am

A really good book that helps you to visualise a mix - I bought this about 5 years ago and it really does open up the visual dimension of perceptualizing a mix.

the art of mixing, a visual guide to mixing

paradigm_x · Post by **paradigm_x** » Wed Apr 12, 2006 1:27 pm

ive found a word doc version of it.

Ive put it up at http://www.paradigmx.com/hipnotic_tut.zip

Written by hipnotic (mod at doa)

cheers

grievous_angel · Post by **grievous_angel** » Sat Apr 15, 2006 8:50 am

KION wrote:After the mixdown process is complete (or during it if you want to check the balance) bounce your track without the limiter first, check both sonically and visually for any spikes by importing your bounced file back into your daw (and line it up with your sequenced track), mute the wav/aiff and play the track.

GREAT piece of advice. And you got it here for free

. If you want confirmation, just take a listen to Kion's tunes on myspace -- the clarity of the mix is excellent.

Sound on Sound just did an interesting comparison of commercial (online) mastering and DIY mastering. DIY didn't do too badly at all.

r33lc4sh · Post by **r33lc4sh** » Sat Apr 15, 2006 9:08 am

KION wrote:And always bounce at 24 bit if you can, so that when you get it mastered/master it yourself, you're less likely to get aliasing on quieter sections with reverb tails going to silence (when the volume is brought up)
Even if you're gonna dither it back down again to 16 bit, the dither noise shaping sounds better than aliasing if bouncing straight to 16 bit before bringing levels up with limiting etc (with a good dither alghorythm). Remember to only dither as the last process, and only dither once (otherwise you start to get artifacts of audible noise)

this arguable - my english is not good enough to explain all the things but i think u'll be able to find enough articels saying that dithering to 16bit can make more shit to your sound than working on a 16bit file from the begining

imho working on higher bit resolution when u are recording electronics which is meant to be played on a big club rig is just waste of diskspace and cpu power

kion · Post by **kion** » Sat Apr 15, 2006 11:42 am

I think you'll find the opposite is true. There's debate about it, but a lot of the conclusions (from reliable sources such as sound on sound) are that you should keep the highest resolution throughout the processs until the very last, where you will dither it down.

I've done tests myself, and found that reverb tails sound far cleaner and free from digital aliasing when the file is kept at 24 bit resolution until the final dither. Dither noise sounds better than digital aliasing

kion · Post by **kion** » Sat Apr 15, 2006 12:02 pm

And this become even more important if you're bouncing tracks down with effects processing within your DAW. If you bounce them at 16 bit, you've instantly shaved off a substantial signal to noise ratio, leaving you with potential quality issues if you have to bounce a piece of audio again. It doesn't actually use up much more hard drive space using 24bit 44khz.

Now if you're talking about using a higher sampling rate, now thats another debatable matter. I don't bother, 24 bit 44.1khz is perfectly fine for me (until the last, when I dither it to 16 bit)

The dubstep tips super thread

Who is online