Ableton Forum

this question is about final mixdown in preparation for mastering - i'd like some feedback on what people do and why ...

and to qualify this - i'm talking about dance music - music for clubs, car stereos and i-pods. so - house, breaks, dnb, techno electro etc ... where volume is definitely an issue and we're not overly concerned with dynamics ... needs to be loud, pumping and defined ..

.. at the same time though, you want to maintain some sense of space around the elements so they can "breathe" .. i dont agree with tracks that look like a straight brick wall in the editor, they just dont sound right ..

there are some people that mix by starting thier kick at 0db, match the bass / lower end with the kick, and move on from there.

samples are nomalized close to 0db. if any compression is used on an element, you use it to raise presence and punch and pull back on the mixer volume for the track to bring it back under 0db and make it sit.

other people swear by normalizing thier samples to a lesser value - say, -3db, then use compression and eq to bring the elements up closer to 0.

which method do you use and why? whats the rational and what are the benefits / drawbacks to both methods ... really curious cause im in the former category and need to finalize a bunch of tracks this summer for release.

.. so either none of you have any idea what you're doing, or you are all running an uninstalled version of live

?
.

What do you do?

i just do what sounds good.

every track is its own and has its own requirements for making different sounds sound good in a mix.

DJRetard wrote:What do you do?

right now i max everything out:

samples are normalized to 0 db the way i record them (with Tapeit)

i base everything around the kick, marry the bass with the kick, then adjust everything around that

if i use compression, i push it till it pops out and sits right, then pull back on the gain on the channel till it sounds right

my results are good imo but i like to know what other people are doing ..

why ... what do *you* do?

(i want to know what *other* people do, i already know what i do )
.

Ive never thought about normalising all my audio but I dont think I would do it for a number of reasons. But doing it probably will up your RMS level which is what you want for the dance stuff. But, you might find you can easily overload the master when mixing the track.

I have no special routine except for what I use on the stereo master. I always use these plugins. Waves Linmb to smooth out the peaks and troughs, then sonalksis compressor to give the track some bite, then I add Voxengo curve eq. This is used to fine tune the whole mix. I usually add two of these. One to perform the EQ duties and the other one for the gear settings which replicate analog boxes. The last in the chain is Voxengo Elephant mastering limiter.


On individual tracks I use whatever I feel like at the time. I do use the same plugins from session to session but nothing esoteric. usual stuff like EQ compression, and I buss tracks a lot.

i set the kick to -6db, then work everything else around it. that leaves plenty of room for some multiband compression on the master bus.

I try to record my stuff at around -6db (not always possible) and only normalise when I have to, I then start with my bass, make it sound good, mix it with the kik, make them sound good together and then start bringing in the other elements, I usually run some gentle multiband compression and a brick wall limiter with a very high threshold (say -2-3 db) on my stereo out, if my track is not close to zero now (which it usually is) I will render and then normalise.

DJRetard wrote:Ive never thought about normalising all my audio but I dont think I would do it for a number of reasons. But doing it probably will up your RMS level which is what you want for the dance stuff.

point/counterpoint: RMS increase will happen using non-linear amplitude scaling, i.e. compression or one of these special (and imho confusingly labeled . . .) RMS normalizers.

Normalizing does does not alter dynamics, it's like a magic fader that auto-sets to the loudest peak to 0db. Only it's not magic

As for normalizing, I always normalize, i guess out of habit and convenience. The convenience part is, when feeding the sound into, say, Live's compressor or eq, it's very easy to conceptualize how many db of effect are happening. Also, these tools have input gain adjustors for the rare case when a normalized signal is too hot for the effect patch.

I can see as how it would be a hassle to normalize with a different workflow though.

Harris.Andrew wrote:

DJRetard wrote:Ive never thought about normalising all my audio but I dont think I would do it for a number of reasons. But doing it probably will up your RMS level which is what you want for the dance stuff.

point/counterpoint: RMS increase will happen using non-linear amplitude scaling, i.e. compression or one of these special (and imho confusingly labeled . . .) RMS normalizers.

Normalizing does does not alter dynamics, it's like a magic fader that auto-sets to the loudest peak to 0db. Only it's not magic

As for normalizing, I always normalize, i guess out of habit and convenience. The convenience part is, when feeding the sound into, say, Live's compressor or eq, it's very easy to conceptualize how many db of effect are happening. Also, these tools have input gain adjustors for the rare case when a normalized signal is too hot for the effect patch.

I can see as how it would be a hassle to normalize with a different workflow though.

I never said normalising alters the dynamics. I know what normalising does.

I hear what your saying when you normalise. But, Imagine the scenario. You have a held lo string note/drone that doesnt have peaks. What benefit are you going to get by normalising something like this. All that will happen here is your compressors will have more input to deal with, aand your faders will have to be very low.

Normalising all aduio in a session is really not nescessary and its only on this forum I have reaad some people do this.

Download the Nine ince nails demo which is a sticky on the main page and youll see what I mean. Check the levels of teh audio, and then normalise everything. See what happens.

Instead of normalising Im using a waves LNMB as the first plug in my master chain. Im getting the same deal as you are but I have a lot more room for my plugins to breathe. Just like in the analog world processors need room to operate efficiently.

DJRetard wrote: Instead of normalising Im using a waves LNMB as the first plug in my master chain. Im getting the same deal as you are but I have a lot more room for my plugins to breathe. Just like in the analog world processors need room to operate efficiently.

word. people jump through hoops for little or no gain ( pun intended) - I love squashing sounds more than most, but maxing out the volume often tends to be a goal in itself for some people, instead of means to an end.

If you normalize realize that you're also raising the noise level.

DJRetard wrote: 1)I hear what your saying when you normalise. But, Imagine the scenario. You have a held lo string note/drone that doesnt have peaks. What benefit are you going to get by normalising something like this. All that will happen here is your compressors will have more input to deal with, aand your faders will have to be very low.

2)Normalising all aduio in a session is really not nescessary and its only on this forum I have reaad some people do this.

3)Download the Nine ince nails demo which is a sticky on the main page and youll see what I mean. Check the levels of teh audio, and then normalise everything. See what happens.

point-by-point, i'm not sure we're on the same page here:

1) first of all, why would i compress something with no dynamic content?

1, alternate answer) what gain would i get from normalize, if i had to se the faders really low - imho, that's what faders are for. this way i can bump it up a few decibals in the mix, or automate a crescendo. and the fader setting will correspond _exactly_ to the the peak amplitude being fed into the mix. to reiterate: i can read and adjust the fader on a channel that's playing a normalized sound with accuracy. i couldn't do this so well if the sound wasn't normalized, i'd have to guestimate how far below 0db the sound is, just gets sloppy imho.

2) actually, you're right it probably isn't. normalizing is easy to do though and i feel like there's benefits: with something below normal gain, i never know how much below normal it really is, so for example setting the threshold on the compressor, just coming up with a baseline number to start with is difficult. Having a good reference number for a baseline really simplifies the (done by ear) fine-tuning. Similar for eq, limiting, gating and the like . . . bread and butter mixing plugs.

3) I haven't d/l'd it, but i imagine it's packaged for garageband users. all the fx / dynamics have been applied already. I guess Reznor wants to let people rearrange easily. Nothing wrong with that, it doesn't preclude more radical reshaping of the sound, just obfuscates the process; better this than putting up sounds that have to be processed by the user in the first place.

DJRetard wrote: 4)Instead of normalising Im using a waves LNMB as the first plug in my master chain. Im getting the same deal as you are but I have a lot more room for my plugins to breathe. Just like in the analog world processors need room to operate efficiently.

4) I don't understand why you say "instead of".

I find using something like LNMB less confusing if you use a normalized signal, as there's a lot of options/knobs that aren't essential in the case of a normalized signal, or that are a lot easier to set when you know you've got a normalized signal coming in.

The functional part of a compressor, the part that actually "compresses" is gain reduction and doesn't need any room to breath. Real world compressors compensate for this gain reduction, but you can set them such that they don't do anything, then . . . yep, normalize.

Machinate: I never ever have to worry about max volume, that's what normalizing does for me.

Mindrise: normalize does raise the noise level, but the signal:noise ratio stays the same, so the noise is in the end no more or less present.

Harris.Andrew wrote: i can read and adjust the fader on a channel that's playing a normalized sound with accuracy. i couldn't do this so well if the sound wasn't normalized, i'd have to guestimate how far below 0db the sound is, just gets sloppy imho.

yes, i agree in the case of individual elements:

record as close to 0db as possible, hottest signal, best signal to noise ratio. or normalize if theyre not already ...

and exactly what you're saying: either way having individual elements (audio files) near 0db simplifies all the other volume/gain controls we have to deal with ... seems the way to go

the original post i made had to do more with final mixdown though ... ok each individual element being near 0db ... where are people setting thier faders and why

ive come to the conclusion that starting with your kick at -5 / -6db is the way to go. why? had a converstation with a friend of mine yesterday and hes been doing it that way for about 10 years ... and the light bulb went off:

you want the final volume of your track to be as close to 0db as possible.

so: you set the master at 0db and dont touch it. then if you start with your volumes on your individual tracks lower, you have headroom to balance them out under and touching 0db

if you start with the kick at zero, you are constantly chasing and pulling everything down - thereby f*cking up your delicate mix - and or limiting at the master and squashing it all when you dont have to and shouldnt be.

made perfect sense to me.
.

Harris,

Normaling audio wasnt really achieveable during the analog tape decades. The rule of thumb was to record as hot as poss. Many in the digital world feel this still holds except theres no tape saturation. The theory behind recording hot iin digital audio is to preserve every bit. I cant say if this is true or not.

Compression


Why would I want to compress a held Lo string/drone.? I can think of a number of scenarios like subbing the kik and lo drone and strapping a compressor across them. While I wont get a sidechaining effect, I will get some kind of pumping effect which I may or may not like.

Electronic music isnt the most dynamic of music but compression is used all the time as an effect, or for a punchy sound. It has nothing to do with the dynamics of the audio.

Im not against normalising audio, but I really see little point in doing it. If some audio is to low I dont normalise. I import in to sound Forge and add gain.

My use of the Waves LNMB as the first plug in my master chain is simply to smooth out the peaks and troughs. This prevents my single channel compressor which is inserted after the LINMB having to work to hard. I suppose you could say this is a form of normalising, but with obvious differences. If I always normaled my audio I could have a hellish time getting the LNMB to operate with enough headroom or any other plugin for that matter. My individual faders would have to be right down as well for the non dynamic audio tracks. (think about that lo held drone and all the other non dynamic sounds I have just normaled to 0db)


Like I said earlier, go and download the nine inch nails track. It gives you a pretty good idea that normaling audio is not nescessary.

But for sample CD's normaling audio is good

If you wish to continue normaling your audio then who am I to disagree. each to their own