What level do you have your mix sit at before mastering?

[Mikaël-ange]

When i was an intern at studio Fredman they actually limited the mix before sending it to mastering haha. THe mastering house mailed and complained but Fredrik just laughed and said that that's how its gonna be

Many big mixer send mix with and w/t limiter to mastering house now... Sound like Fredrik was a visionary

 

[solphilcox]

never really found anything wrong with (gentle) limiting on an unmastered mix. i generally just put the waves SSL console comp on the master bus and try to keep any peaks under a few db or so. never heard of a mastering engineer complain that a mix hasn't been bounced loud enough, but i see how slamming it to hell can cause problems.

 

[Mikaël-ange]

never really found anything wrong with (gentle) limiting on an unmastered mix. i generally just put the waves SSL console comp on the master bus and try to keep any peaks under a few db or so. never heard of a mastering engineer complain that a mix hasn't been bounced loud enough, but i see how slamming it to hell can cause problems.

Unless you use 10:1 ratio with waves ssl that not limiting but compression.

 

[crillemannen]

never really found anything wrong with (gentle) limiting on an unmastered mix. i generally just put the waves SSL console comp on the master bus and try to keep any peaks under a few db or so. never heard of a mastering engineer complain that a mix hasn't been bounced loud enough, but i see how slamming it to hell can cause problems.

Yeah that is not limiting that is compression...

 

[Eduardo Apolonia]

I'll explain why I use -0.2 dBFS:

This is my usual mix chain:

1 - Pultec
2 - Shadow Hills Mastering Compressor/SSL 384/API 2500/Focusrite 330
3 - Crane Song Ibis
4 - Crane Song STC-8
5 - Crane Song HEDD with processing

I like to clip the HEDD a bit on the way in. It actually does a kind of limiting because it rebuilds the clipped peaks so they don't end as flat clips and gives an extra "smack" to the mix.
The HEDD max digital output with the processing on is -0.2dBFS so that's why I use that as my max peak.
There's no headroom but the mastering engineer can lower the volume digitally before hitting is mastering chain if he needs.

My mixes are about 2-5 dBs RMS lower than the final result.

I never agreed in delivering to the mastering engineer mixes that have a dynamic range of 8-12bBs to be squashed. That could make all the bounces that were so hard to achieve go out of whack.

 

[Diego Rugiero]

I think peaking around -10dBFS works for all master purposes.

 

[Uros]

I'll explain why I use -0.2 dBFS:

This is my usual mix chain:

1 - Pultec
2 - Shadow Hills Mastering Compressor/SSL 384/API 2500/Focusrite 330
3 - Crane Song Ibis
4 - Crane Song STC-8
5 - Crane Song HEDD with processing

I like to clip the HEDD a bit on the way in. It actually does a kind of limiting because it rebuilds the clipped peaks so they don't end as flat clips and gives an extra "smack" to the mix.
The HEDD max digital output with the processing on is -0.2dBFS so that's why I use that as my max peak.
There's no headroom but the mastering engineer can lower the volume digitally before hitting is mastering chain if he needs.

My mixes are about 2-5 dBs RMS lower than the final result.

I never agreed in delivering to the mastering engineer mixes that have a dynamic range of 8-12bBs to be squashed. That could make all the bounces that were so hard to achieve go out of whack.

You do have a point in minimizing the ME's chance to fuck the mix up.
But if you work along with a ME which you know well, these precautions are not necessary.

 

[NickL]

Here's what Bob Ludwig told me:

Use up all the bits for best dynamics resolution. Even a clip light here and there is fine as long as it's short and non-audible, but make sure to use up all the bits.

I would not argue with Bob Ludwig if you paid me

Edit: This does NOT, however, mean that you should limit the shit out of everything to get it as hot as you can.

 

[TheDustbowl]

Actual levels are not the whole story. I can produce a mix with a mean RMS level of -18 dBfs that is identical to one at -3 dBfs except for level - so long as the difference between the mean RMS and the highest peak is less than 3dB in either case. The point of staging your gain structure is to 1) hit your processors with a level that they will respond well to (this is massively debated regarding plugins (I do it) and not at all reasonably debatable regarding analogue outboard) and 2) to allow for dynamics - the region above your mean level is your headroom, this is where the transients and other peaks are allowed to breathe and give your music dynamic punch. Your ME is the one to reduce this level in order to get as much level as possible (yup, it's the real world here) without destroying the impact of the music (too much) - BUT - if there is no punch - if there are no peaks in the mix - then it really doesn't matter - you may have already killed it. The key measurements here are crest factor (peak divided by mean) or dynamic range (peak minus mean) which gives us an idea of the difference between mean and peak level.

To put some numbers to it, I'll happily mix to -18 or -20 dBfs RMS with a DR of around 14 which gives peaks of -6 to -4. Bear in mind that some music simply doesn't have this much dynamic range to start with, and some needs far more. I wouldn't get too hung-up on the exact numbers, just leave some breathing space.

Bob Katz and Ian Shepherd are worth a Google if you want to go into more detail.

I'll happily squash a track or 2 for you if you want to hear what happens to them.

Regards to all

A.

 

[TheDustbowl]

Here's what Bob Ludwig told me:

Use up all the bits for best dynamics resolution. Even a clip light here and there is fine as long as it's short and non-audible, but make sure to use up all the bits.

I would not argue with Bob Ludwig if you paid me

Edit: This does NOT, however, mean that you should limit the shit out of everything to get it as hot as you can.

Are you sure this is in context? 16 bits gives 96dB of dynamic range, 24 bits gives 144dB and 32 bit float gives 1680dB. You should be handing mixes to the ME at the highest bit-rate you can work at - in most DAWs that is 32bit float. Did Bob Ludwig REALLY suggest that you need 1680dB of dynamic range? Could he have been suggesting that the mastered peaks can occasionally clip, not the mixed tracks? This isn't dynamic resolution, it's just loudness!

You can fit everything from the quietest that you will ever hear to the loudest that you could stand in 20 bits, you do not ever need to go over for dynamic range.

However, there IS a reason that some engineers will do it - if you square a signal peak it will cause a DA converter to produce a "fake" transient which means that you can actually make the RMS louder because you cheat the 0dB limit for your peaks.

A.

 

[crillemannen]

With todays mastering climat i think it is important to atleast mix with a limiter so you don't get any ugly surprises after you have send it to mastering. Levels on instruments can be fukked up if you mix at -18 and don't check your mixes when it hits the 0. But as any DAW its not important how hot your mixes are (or cold) just don't let it hit the 0 to be sure.

And for todays modern mixes with the music kind of based around the snare you should be fine if yo just make the snare to just reach yellow on the tracks which usually is -18 then your whole mix should end up around -10 to -5. I usually use the plugin TRIM first on all of my tracks so all the faders are at 0 and then only use the faders to automate things later on.

 

[waltz mastering]

Here's what Bob Ludwig told me:

Use up all the bits for best dynamics resolution. Even a clip light here and there is fine as long as it's short and non-audible, but make sure to use up all the bits.

I would not argue with Bob Ludwig if you paid me

Edit: This does NOT, however, mean that you should limit the shit out of everything to get it as hot as you can.

Was that a long time ago? or recently? When older 16 bit recording audio formats used to be in vogue (adat , d88, PT), that used to be the mantra .. it's not really an issue with 24 bit.

In other words,.. I don't think you'll find to many if any pros that would agree with that today, where they would have 15 or 20 years ago.

 

[NickL]

Are you sure this is in context? 16 bits gives 96dB of dynamic range, 24 bits gives 144dB and 32 bit float gives 1680dB. You should be handing mixes to the ME at the highest bit-rate you can work at - in most DAWs that is 32bit float. Did Bob Ludwig REALLY suggest that you need 1680dB of dynamic range? Could he have been suggesting that the mastered peaks can occasionally clip, not the mixed tracks? This isn't dynamic resolution, it's just loudness!

You can fit everything from the quietest that you will ever hear to the loudest that you could stand in 20 bits, you do not ever need to go over for dynamic range.

However, there IS a reason that some engineers will do it - if you square a signal peak it will cause a DA converter to produce a "fake" transient which means that you can actually make the RMS louder because you cheat the 0dB limit for your peaks.

A.

Hi there, for what it's worth I was asking Mr. Ludwig about a session that was recorded in 24 bit, not 32. And since it was asked, it was fairly recently (summer 2011) Perhaps its ignorant of me to assume that most people here are using 24 bit, but of the various studios I've worked in/assisted in I believe only one was setup for 32 bit floating point so I simply did not consider it.

However, as Waltz pointed out, 24 bit is generally considered to be the point at which this method fails to hold water for many in the know on digital audio. Bob Katz I believe has stated that one should leave at least 3db of headroom for intersample peaks. But I have heard other stories of people being told by Mr. Ludwig to aim for 0dbfs on peaks. Perhaps his digital audio theory is becoming somewhat antiquated? I would be surprised but its certainly possible...

But to my ears he's the best in the business and if he feels like mixes which approach 0 db sound perfectly acceptable then I am inclined to trust his judgement, not based on any particular technical grounds. You will hear lots of different suggestions out there for appropriate mix levels and some guys still suggest VU meters, which can be slow enough to allow 20dbvu peaks to slip by unnoticed on quick transients! To hear one of the worlds top mastering engineers recommend peaking near zero says to me that this is not as critical an argument as it may seem, just keep your peak meter out of the red and have fun mixing. I do not consider myself a mastering engineer so feel free to set me straight if you have to. Cheers.

Edit: I really wish I could get ahold of him to receive and share some clarification on his thinking, but the last time I inquired about hiring him I was told I'd have to settle for his former assistant Adam Ayan and couldn't get through to him haha

 

[TheDustbowl]

32 bit FPA has been common in non-PT rigs for years.

Bob Ludwig is undoubtedly one of the best and he has produced some masters and re-masters that are both very hot and very good sounding, but in the context of this thread the suggestion that all of the bits need to be used for maximim dynamic resolution is wrong.

If you are considering purely the maths then yes, but the filter of human hearing totally invalidates the argument unless you are dealing with low wordlength systems, which there is no reason to do in the second decade of the 21st century.


A

 

[NickL]

32 bit FPA has been common in non-PT rigs for years.

Bob Ludwig is undoubtedly one of the best and he has produced some masters and re-masters that are both very hot and very good sounding, but in the context of this thread the suggestion that all of the bits need to be used for maximim dynamic resolution is wrong.

If you are considering purely the maths then yes, but the filter of human hearing totally invalidates the argument unless you are dealing with low wordlength systems, which there is no reason to do in the second decade of the 21st century.


A

I'd be interested to see a poll on how many people here are using 32 bit. As a studio guy I rarely run into non-PT users but you make a valid point, a lot of people here are using cubase/sonar/logic/etc but I wonder how many are actually using 32 bit resolution. For these people I agree with you, full use of bits is totally unnecessary and leaving some extra headroom for intersample clipping would be beneficial. I also wonder what percent of mixdowns that are sent to mastering houses are in 32-bit formats... But I don't think one needs to worry themselves too much about RMS levels during mixing unless the mastering engineer requests something specific, and I concede that this probably includes trying to get the levels as close to zero as possible (definitely in 32 bit). Maybe Mr. Ludwig feels his processing equipment responds better to signals that are recorded with all of the bits utilized? Wish I could ask him...
On a side note, I have been told that in Pro Tools there is some additional headroom above the clip light threshold as a safety feature. I don't know how accurate this is but I have indeed seen mixes with one or two clip lights that had no audible distortion whatsoever. I, personally, would not send off a mix with any clip lights on the master fader but I'm just recounting what I've been told. I always try to keep in mind that there's a lot of opposing ideas and misinformation out there.
Once again, keep it safely out of the red and have fun, or as you said earlier, don't get too hung up on exact numbers. Agreed?

 

[TheDustbowl]

PTHD used to run a 48 bit FP mixer so not too many worries there. For anyone who mixes and sends tracks to a Mastering Engineer - mix-down at full 32 or 48 bit and send the 2 track at that resolution; if you're mixing at anything less then stop doing it. Note that this is NOT the same as recording at 24 bit - the moment you do ANYTHING to any track in your mix you will (almost certainly) make wordlengths longer; record at 24 bit, mix at 32 or 48 and pass the files to your ME at full mix resolution. Let the ME work the full wordlength and then reduce and dither at the end.

NickL - totally - record at 24 bit, mix at 32 or 48 and then don't worry, the dynamic range will easily swallow all of human hearing pretty well whatever you do - obviously if you're mixing for bats with hearing aids then YMMV

A.

 

[NickL]

PTHD used to run a 48 bit FP mixer so not too many worries there. For anyone who mixes and sends tracks to a Mastering Engineer - mix-down at full 32 or 48 bit and send the 2 track at that resolution; if you're mixing at anything less then stop doing it. Note that this is NOT the same as recording at 24 bit - the moment you do ANYTHING to any track in your mix you will (almost certainly) make wordlengths longer; record at 24 bit, mix at 32 or 48 and pass the files to your ME at full mix resolution. Let the ME work the full wordlength and then reduce and dither at the end.

NickL - totally - record at 24 bit, mix at 32 or 48 and then don't worry, the dynamic range will easily swallow all of human hearing pretty well whatever you do - obviously if you're mixing for bats with hearing aids then YMMV

A.

Dolphins with oscilloscopes actually

 

[TheDustbowl]