A guide to home "Mastering" of your own tunes

[SigEnt]

Just dropping a thank you in this thread. This guide has been a massive help in getting my tracks to a decent volume for soundclound and so on, and I don't get that "Oh shit" feeling when it's time to master my tracks.

Try hitting youtube on the subject of mastering and you will end up very confused quickly - the hobbyist like me just wants to get their tracks to a decent level so we can put 'em on sound cloud without totally wrecking them.

Big thanks again.

[Mattias]

Briefly read it through. Seems great.

Not sure what / how autonormalization works in Ableton but I assume it's non destructive within the project file.

[Mattias]

Before judging, why would you want to do that?

[Lost to the Void]

Well, to cut a very long story short.

The arbitrary -3, -5 etc peak that we mastering engineers ask for is basically a safe guide.
To ensure people don`t clip.
Generally if I have a regular client or someone who I know knows the do, then I will work with any peak level as I know there will be no clipping.

Bare in mind a lot of meters don`t account for TruePeak, and I`m not sure how reliable abletons normalisation is, but there is absolutely no reason to normalise to send to your engineer. They don`t give a shit how loud the final is, you won`t be impressing them, they are looking for clipping, and overall dynamic range.

So as a general guide -3db is a pretty safe margin, but basically just make sure you don`t clip.

[Mattias]

You want to be looking at dBTP though. Also you wont win any "maximize bit rate" during normalizing (I don't get what you would want to do this anyway? Sounds like you're buying into bad internet audio ideas) your audio like that.
I would say it's bad practice, that eventually could be harmful depending on the accuracy of the DAW like Steve mentioned. Never do anything if there is no particular reason for it, which, in this case, there isn't it.

[Lost to the Void]

I think, personally, it's better overall to improve your monitoring and working levels so that you produce to a consistent level.
I think I mentioned this earlier in this thread.

But if you aim to produce your music ways to the same RMS level and monitor at the same level your mix decisions will become easier and better.

So for example..

When. I produce Techno, or most types of dance music, I tend to work to a peak of -1 ish (I use True Peak metering so I am safe) and an RMS around -12 or -10
That's loud enough that none of my monitoring gain needs to be pushed hard to get a decent listening level, and it's easily "listenable" without needing to do a test master if I want to check it out and about.
Also it's been controlled dynamically so the mastering engineer doesn't have to work too hard and there won't be any surprises, dynamically speaking, when it comes back from the engineer.

Whatever you are comfortable with, I can get this level without crushing the shit out of things.

You don't need huge dynamic range in dance music, it's utterly wasted. And there is no reason to be producing with crazy low RMS like -18 or -24.

Some techno guys send me stuff and it's sooooo quiet, it baffles me how they can get any decent volume out of their monitoring.

If you are consistent with your production levels and monitoring there is no need for normalisation, and you will develop better techniques regarding dynamics and mix balancing.

It does take a while to get this kind of consistency, you need to think about dynamics, but it's something you should be aiming for.

[Lost to the Void]

I think Matty understands normalisation. I think what he is getting at (same as me) is that we are unsure if it works on TruePeak (a rather new standard) or just flat out digital 0db as the "limit".

Normalisation doesn`t do anything to dynamics (apart from potential clipping due to ISP/TruePeak not being conformed to)

If your music has a DR of 8db, and peaks at -6, if you normalise it, it will still have a DR of 8db, but it will peak at 0.

So you won`t be making the most (or less) of the dynamics, and ITB noise floor shouldn`t be an issue.

[Lost to the Void]

TruePeak is about DtoA reconstruction, it's effectively about looking ahead to future processes.

https://www.masteringthemix.com/blogs/l ... k-metering

I'm not sure how autonormalisation takes this into account, or if it does.

In fact I can't find any documentation on the autonormalisation process but I fail to see how it could increase your dynamics. Actually if it does, and it is acting like an expander and increases the dynamic range of a track then this is a bad thing as you want your dynamics to be.... Well, whatever you decide them to be, and not something that gets altered after rendering.

If you can link me to any documentation that would be great.

[Mattias]

I dont get the "increase dynamics" bit, why would autonormalization do that?

Is it applied pre or post fader? It definitely will clip if you feed a normalized signal into plugins that you process the audio with etc.

I dont have Ableton myself so I cannot run tests what it actually does myself.

But, I may be wrong here, what I gather and understand from your explanation (correct me if I'm wrong), you want to have the best possible "outcome" of your bit depth right? Thing is, at 24 bit you have loads of headroom and very very little noise to begin with. If you capture at 16 bit and use that throughout the project I would still be careful since normalization raises the noise-floor as well as everything else. Finally, PCM is always full bits. So to me, no benefit at all.

[Lost to the Void]

So it's basically the same as offline normalisation, there's no indication from that text it has any effect on the dynamic range (which it shouldn't do anyway)
I'll write to them for clarification to see if it accounts for ISP/TruePeak
That none of their own metering accounts for this makes me doubtful, peaking at zero on Ableton is dangerous using their own meters as you will get TruePeak readings of up to +3db.
Either way I see very little point unless you are producing with your peaks at some silly low order like peaking at -12 or something.
Normalise would make sense if you could set the output level. (Which you could do at -0.5 dB or something, which is what most of us engineers do for digital destination masters)

[Mattias]

Yeah no normalization shouldn't affect dynamic range ever. I mean the whole point of normalization is to increase the same values, they will be left as they were but louder.

edit spelling

[Lost to the Void]

Ok, done as much reading on this as I could find, and essentially, as I thought, normalisation is pointless when talking about a whole mix, unless your levels are ridic low (in which case I referred to above a page back or so, for good general monitoring/level practice, which removes the need for normalisation).
Normalising downwards can lead to quantisation error, but in this discussion we are talking about normalising upwards, which will not cause quantisation error, nor will it effect dynamics (unless ableton have done something drastically stupid with their coding, which I will find out).
As we are unsure how or even IF ableton normalisation accounts for ISP/TruePeak, I wouldn`t do it at all.

Basically it`s not necessary, as there is no benefit.

[Mattias]

Splendid of you to read up on that to confirm what we suspected from the beginning. Cheers for the effort.

Also just noticed the forum status of Steve "Shit Lord" and mine "Has Shit His Pants" hahah.

[Lost to the Void]

Haha.
It's actually pretty hard to find anyone who really knows the deal on the mechanics of Ableton
I've found what I can and also written in to the Devs, so hopefully I'll get technical clarification.

I find it hard to reason that Ableton has done something out of the ordinary with normalisation, especially regarding dynamics, that would be incredibly stupid if they have.

[juodas]

Well i recognized that, as soon as i listen bounced version a volume levels of it is way to low compare when i'm a listening on a daw i'm always trying to work between -6 : -12 db, never used normalisation, however i think i will start using it...

[Lost to the Void]

Why?

[0dd]

In order to increase dynamic range it would need to only affect peaks no? And that would then be expansion, not normalisation.

[Lost to the Void]

Well exactly. I would be extremely worried if my carefully set dynamic range was altered by rendering with absolutely no input control from me.

[juodas]

Well exactly. I would be extremely worried if my carefully set dynamic range was altered by rendering with absolutely no input control from me.

What do you mean by that, should i contact ableton or reinstall the software, well i started to worry about the errors and crashes in ableton for no fucking reason

[Mattias]

In order to increase dynamic range it would need to only affect peaks no? And that would then be expansion, not normalisation.

I don't follow. Why would it not work if it affects every sample?

Because dynamic range is the relation between the highest and lowest events. If everything is increased equally, like when one uses normalization, the dynamic range stays intact. It was how it used to be pre-processing so the normalization was just something unnecessary like we stated previously.

No true need for you to continue normalizing your audio gents