You are here

Premaster volumes, rnb singers.

All about the tools and techniques involved in capturing sound, in the studio or on location.

Premaster volumes, rnb singers.

Postby SheldonTrever » Wed Oct 16, 2019 7:09 am

Hello peeps,

I need an opinion, I’m doing a mix/master for one of my first clients and I’m almost ready to master but going into my master is there a prime volume or db I should be at before even starting mastering? The main mix is hitting-3, -6 and 0 on some parts Any tips or advice?
SheldonTrever
Posts: 1
Joined: Wed Oct 16, 2019 7:04 am

Re: Premaster volumes, rnb singers.

Postby Wonks » Wed Oct 16, 2019 10:13 am

You are way too high!

Pre-master, you should be aiming for an average of around -18dBFS, which means turning all the channel levels down to achieve that, not the master channel faders. And that means proper gain staging at the start of each channel, not just the channel faders. So you'll need to either use the channel's input gain control, or add in a gain module if your DAW doesn't have an input gain control.

This will also mean adjusting the settings on any insert compressors you have.

And when mastering, you need to keep below a true peak value of at least -1dB. Never, ever get to 0dbFS.
User avatar
Wonks
Jedi Poster
Posts: 9728
Joined: Thu May 29, 2003 12:00 am
Location: Reading, UK
Correcting mistakes on the internet since 1853

Re: Premaster volumes, rnb singers.

Postby The Elf » Wed Oct 16, 2019 10:48 am

Wonks wrote:that means proper gain staging at the start of each channel, not just the channel faders.
+1

A good starting position would be to adjust all your gains to achieve max -10dBFS peak per channel (that's just my personal guide - many say -18dBFS average) with faders at unity.

And, as I always say, if you find you're having to touch the master faders - even though many will tell you this does no harm - you've lost control of your mix.
User avatar
The Elf
Jedi Poster
Posts: 12867
Joined: Tue Aug 14, 2001 12:00 am
Location: Sheffield, UK
An Eagle for an Emperor, A Kestrel for a Knave.

Re: Premaster volumes, rnb singers.

Postby Wonks » Wed Oct 16, 2019 10:54 am

I was saying around -18dBFS average (RMS value on the meters), so peaks would be higher (e.g. 10dBFS). And that's at the loudest part of the recording, so look at the waveform and select a section where it looks the loudest.
User avatar
Wonks
Jedi Poster
Posts: 9728
Joined: Thu May 29, 2003 12:00 am
Location: Reading, UK
Correcting mistakes on the internet since 1853

Re: Premaster volumes, rnb singers.

Postby Dr Huge Longjohns » Wed Oct 16, 2019 10:54 am

And when mastering, you need to keep below a true peak value of at least -1dB

Is this what most mastering engineers use for delivering tracks? I was under the impression it was more like 0.1dB on the limiter these days?
User avatar
Dr Huge Longjohns
Frequent Poster (Level2)
Posts: 3217
Joined: Thu Apr 10, 2003 12:00 am
Location: Gallifrey
"The performance is 99.9% of what people hear"- J. Leckie
"It's all complete nonsense, anyone who knows what they're doing can deliver great results with whatever comes to hand" - H. Robjohns

Re: Premaster volumes, rnb singers.

Postby Hugh Robjohns » Wed Oct 16, 2019 10:58 am

The Elf wrote:A good starting position would be to adjust all your gains to achieve max -10dBFS per channel (that's just my personal guide - many say -18dBFS) with faders at unity.

I concur! :-D

I'd recommend having the signal hover mostly (ie average) around the -20 to -18dBFS area, with transient peaks to no more than -10dBFS.

In the previous age of 'peak normalisation' mastering engineers would have to make the signal hit the 0dBFS endstops, and often they ended up with inter-sample peaks going higher, either accidentally or intentionally! This was not big or clever, and everyone knows it sounds crap... but the lunatics ran that particular asylum...

These days, where pretty much all the main streaming services and broadcasters have seen sense and are using 'loudness normalisation', it is the integrated loudness value that matters most -- usually between -16 and -12LUFS -- and peaks should be no higher than -1dBTP ever, and some require -3dBTP (to allow processing headroom for lossy codecs).

H
User avatar
Hugh Robjohns
Moderator
Posts: 24781
Joined: Fri Jul 25, 2003 12:00 am
Location: Worcestershire, UK
Technical Editor, Sound On Sound

Re: Premaster volumes, rnb singers.

Postby The Elf » Wed Oct 16, 2019 11:09 am

Wonks wrote:I was saying around -18dBFS average (RMS value on the meters), so peaks would be higher (e.g. 10dBFS). And that's at the loudest part of the recording, so look at the waveform and select a section where it looks the loudest.
Yeah, sorry - I meant to say '-18dBFS average' - I use peak, rather than average as my guide. I've amended my post.
User avatar
The Elf
Jedi Poster
Posts: 12867
Joined: Tue Aug 14, 2001 12:00 am
Location: Sheffield, UK
An Eagle for an Emperor, A Kestrel for a Knave.

Re: Premaster volumes, rnb singers.

Postby Dr Huge Longjohns » Wed Oct 16, 2019 11:12 am

peaks should be no higher than -1dBTP ever, and some require -3dBTP (to allow processing headroom for lossy codecs).

OK, I've been living in the past clearly! Thanks.
User avatar
Dr Huge Longjohns
Frequent Poster (Level2)
Posts: 3217
Joined: Thu Apr 10, 2003 12:00 am
Location: Gallifrey
"The performance is 99.9% of what people hear"- J. Leckie
"It's all complete nonsense, anyone who knows what they're doing can deliver great results with whatever comes to hand" - H. Robjohns

Re: Premaster volumes, rnb singers.

Postby Hugh Robjohns » Wed Oct 16, 2019 11:33 am

It was the fashion to hit the digital sample-peak meter endstops regularly in an effort to make the latest track 'louder' than everyone elses...

Some engineers, thinking digital clipping was a bad thing (it generally is!) decided they could assuage their conscience by limiting to -0.1dBFS so it wasn't quite hitting the end stops...

Except that standard sample-peak meters only show the highest amplitude of a group of digital samples, and not the actual amplitude of the reconstructed (analogue) waveform. Unfortunately, it is quite possible for the reconstructed amplitude between digital samples to be higher -- by as much as 9dB in really extreme cases, and 3dB quite commonly -- and this can overload the D-A converter or a lossy codec (such as when making an MP3 conversion).

These are called inter-sample peaks.

So, as part of the loudness normalisation suite of tools, there is a 'true peak' meter which works by oversampling the digital signal to get a much better idea of what's going on between the original samples. True-peak meters display level calibrated in dBTP -- the TP identifying the true-peak nature of the measurement, and usually go up to +3 or +6dBTP so that they can accurately register inter-sample peaks in peak-normalised material.

Most true-peak meters use 4x oversampling, which gives a potential error in the peak amplitude of 0.688dB. Higher oversampling rates are more accurate but impose a higher processing overhead, so 4x is the standard and, to take into account the possible 0.688dB under-read, the rules are that no true-peak shall ever exceed -1dBTP, thus absolutely guaranteeing that the reconstructed signal will never exceed 0dBFS under any circumstances.

A later recommendation is that the true-peak value shouldn't exceed -3dBTP if the signal is likely to be processed by a lossy codec (eg. MP3 etc) because the signal processing involved requires more than 1dB of headroom!

So... if you're uploading material to a streaming website, I'd recommend mastering so that the true-peaks are between -3 and -1dBTP, and the integrated loudness is around -12 to -16LUFS, depending on the specific platform. -14LUFS seems a practical compromise value!

H
User avatar
Hugh Robjohns
Moderator
Posts: 24781
Joined: Fri Jul 25, 2003 12:00 am
Location: Worcestershire, UK
Technical Editor, Sound On Sound

Re: Premaster volumes, rnb singers.

Postby CS70 » Wed Oct 16, 2019 1:14 pm

SheldonTrever wrote:Hello peeps,

I need an opinion, I’m doing a mix/master for one of my first clients and I’m almost ready to master but going into my master is there a prime volume or db I should be at before even starting mastering? The main mix is hitting-3, -6 and 0 on some parts Any tips or advice?

Absolutely - the bottom line is "lower the master fader to give yourself more working headroom during mastering". Most mastering engineers I've worked with, ask for -6dBFS with a minimum of -3dBFS of headroom on the 24 bits mix.

To add to what others have said: it's a critical practice to record at sensible levels, since the A/D converters will usually work at best well into their performance envelope. Recording too hot typically ends up producing more quantization errors in the sample stream and thus worse, less detailed sound even if you don't hit digital distortion.

For the similar reason that it's gonna be D/A converted on playback, your final 24 bits mix should peak as above (i.e. well within the D/A converter dynamic range).

However, once you have a recorded signal (i.e. a sample stream) and you are into the DAW, keeping levels down it's still very good practice but - with some qualifications - not crazy necessary.

That's because internally DAWs work at 32 or 64 bits and "signal processing" is simply a set of calculations designed to work in a finite 32 or 64 bits number space.

Working with so many bits makes so that each processing stage has enormous headroom.
Stages being stuff like the virtual "channel strips" with their built in EQ, sends/returns and of course the virtual "mixing circuit" itself, aka "the calculations which sum up the channel sample streams". They don't have infinite headroom (you'd need real real numbers for that :) ) but in practice it makes little difference.

This is very different to analog mixers and hardware effects, which usually have a good dollop of safety headroom on each channel and input, but can be overloaded if you really do something stupid (or drive that 1176 into distortion because it sounds good).

For all practical purposes, you can't overload a DAW channel, at least with signals coming from an audible analog recording. But - I repeat - you can certainly overload the D/A converters if your 24 bit master bus exceeds the converter's dynamic range... so your final mix must not be too hot.

Also your final 16 bit master must have sensible levels since, unlike DAWs, cd players or "normal" software music players cannot be expected to add bits to each sample to provide more headroom. They will take whatever you give them and try to D/A convert it.

As for plugins, it depends a little on their architecture. Some maybe coded to resemble analog headroom behavior and thus have an artificially limited headroom (purposely distorting samples over a certain number, or using calculations which may produce errors if the sample values are outside the "analog-like" range); some others may be designed like that by mistake, unintentionally or as a result of poor testing. Many plugins however handle 64 bits samples with no problems and happily process them correctly, and pass the back to the DAW mixing engine without fuss.

The consequence of all this is that if you like the sound you hear (meaning that no plugin is distorting your signal), but your master bus is too hot, you don't really need to go back to each channel and bring it down, but it's enough to lower the master bus fader. The DAW's crazy big headroom will work in your favor and produce results (i.e. calculations) which are as correct as if you had sensible (24 bits) levels in every channel.

Note that this is only an emergency thing: when you have already sweated on a mix using "wrong" levels (because you didn't know) but nevertheless you like the sound of it, and don't want to fiddle with the balance.. since in complex mixes with post-fader sends, changing the fader levels may change the timbre of the result.

In that case you can safely "simply" dial back the master bus fader to be able to bounce a mix which can be then mastered.

(All this obviously it doesn't work if your DAW uses 24 bits internally... but nowadays most don't).

Once you know, however, much better to stick with not too hot levels in the first place!
User avatar
CS70
Jedi Poster
Posts: 4341
Joined: Mon Nov 26, 2012 1:00 am
Location: Oslo, Norway
Silver Spoon - Check out our latest video  and the FB page


Who is online

Users browsing this forum: No registered users