Questions about master levels in classical music...insight needed

[Manozi]

Hey all,

A while back I did my first serious on-location classical recording for a chamber music festival in Norway. I am now in the process of mixing the tracks and I have to send them back to the festival soon.

Since I have never mixed a classical music project that will be released on a 'commercial' CD, I have been wondering about a number of things.

Mostly I have questions regarding the levels for the final stereo track, so I thought I'd post this here since it is in a way related to mastering.

In a nutshell I can't figure out where to have the final levels for the tracks.
Obviously everything shouldn't be slammed since dynamics have to be preserved to reflect the nuances of the style, but just how 'loud' should the final mix be here?

Should the final stereo track be normalized? Or should I just get a good mix and leave it alone even though it peaks somewhere around -8.0Db (like some of my mixes do at the moment)?

I know in cases like this it's good to use references and see how others work. In this case I am using two reference recordings - one from the other engineer (who recorded some of the performances from the same festival) and one from the festival's official CD from 2005.

The final mixes on the 2005 disc are all very low in level, peaking somewhere around -8.0 - -7.0 Db. Initially I went with this as my reference so I had my levels in the same ballpark.

Later on however I got to listen to the other engineer's mixes and it seems like he normalized everything to the peak values (loudest parts of the mix).

I am trying to figure out which way to go with these. My instinct tells me to go with the normalization, but I am not entirely sure since I have relatively little experience in dealing with this style.

I also figured that if both the other engineer's and my mixes are to be released on the same disc, they should be identical in level from track to track. And no, this is not going to be mastered, so level adjustments cannot be applied at this stage.

Now for the questions:

1) Is there a difference between pushing up all the faders a bit more in Pro Tools to get a better master level when doing the final mix, and applying 'Normalize' in Sound Forge to a completed mix (stereo track) to accomplish the same thing - bring the master levels up? (This is the software I'm working with so I'm using it in the examples) Would one sound different than the other?

2) This is more of a mixing question, I suppose, but it is something I have been wondering about for a while, especially while mixing these.

Once I have a good sounding mix, and fader levels on all tracks are set in Pro Tools, can I turn all the tracks up in level together while preserving their relative levels to one another without affecting the mix itself? For example if I create an 'All tracks' group and boost the whole group up in level. Does that affect the mix?

3) Does Normalization applied to the stereo track after the mix is completed affect the mix itself? I.e. - are my mixes going to sound different after I normalize them? I don't want to ruin the mix just by tweaking the levels.

4) Again, I guess more of a mixing question but - what processing (if any) is applied to the master bus when mixing classical material?

My approach is the purist one - I don't use any processing or effects anywhere (including the master bus) unless really needed. I am setting the levels and panning and leaving it at that. But I have been wondering - is it a good idea to use compression etc. on the master? Obviously the Waves L2/L3 would not be a good idea, but is the concept of compression on the master bus in classical applications relevant at all? Is it used? And if yes, what are some analog (and plug-in) units that work for that.


I have to submit the final mixes soon, and I have to figure this out, so I would appreciate everyone's help!

Any and all comments are appreciated and will be useful



Thanks in advance!

-Manol

[reynaud]

3) Does Normalization applied to the stereo track after the mix is completed affect the mix itself? I.e. - are my mixes going to sound different after I normalize them? I don't want to ruin the mix just by tweaking the levels.

I wouldn't normalise the material. A better approach is to raise the overall level in your DAW and let the material peak at around -5dBFS (With some material recorded by other engineers where the peaks are hitting 0dBFS and where any processing is required I have been known to use a safety limiter but in your case this is not a problem). The previous engineer's material can also be raised in the DAW to match your levels. Of course its always the case of listen and see if you hear any artifacts or if it doesn't sound good or suit the material. It always comes down to achieving the producer's or the performer's goal. Your job is to match that as close as possible.

Generally when recording I try to have the peaks in this range so as little processing is required after the fact. And my goal is always to capture the performance as faithfully as possible using good mic technique in a good hall or room so as to minimise the need to process in the DAW. But sometimes you've got to compromise a little to meet the goals of all involved.

4) Again, I guess more of a mixing question but - what processing (if any) is applied to the master bus when mixing classical material?

As little as possible and only if really necessary. Usually the producer decides if processing is needed and if any editing is required (at which point source-destination editing with a good crossfade editor makes life simpler and much faster). There is nothing wrong with using light compression to tame some dynamics especially with overly dynamic music where it may spoil the listening experience. But with chamber music this is normally not a problem. Generally, preserve as much dynamics as possible. Parallel Compression can sometimes be a solution to overly dynamic material or you could use "bottoms up" compression where only the softer parts are brought up and the loud parts are left untouched.

For best results use the Weiss DS-1MKII digital compressor. Flux Solera compressor (RTAS) plugin can also deliver good results if used appropriately. PSP's MasterComp (RTAS) in small doses can also do the trick. The goal is always to get and in and get out without being noticed when dealing with classical material especially if the producer is attending the session.

In terms of EQ, the only time to ever really use an EQ is to correct bad mic placement (at which point you know you have done something wrong) or to lessen the effect of rumbling on the lower octave. Again get in, do your thing and get out without affecting the material too much. Here the Weiss EQ1-LP Digital EQ or the Algorithmix Red or Orange plugin EQs (PC VST or DirectX only) produce fantastic results. If on a budget you could use PSP's Neon HR Linear Phase EQ (RTAS), while not as good as the Weiss or the Algorithmix, it can also produce good results.

hope that helps.

cheers,
Reynaud

[Thomas W. Bethe]

I think that reynaud's post sums things up very nicely.

For classical I try and stay in the -6 dBFS to maybe a peak at -3 dBFS range and let the rest of the music lie where it is. I try never to use ANY compression or limiting or normalizing but try and do the recording in such a way that I don't have to play with the levels to much in post production and if I do I use the "gain" adjustment in WL for increasing the gain.

We do a lot of amateur groups and the levels can be all over the place and I may, in that situation, keep my recording peaks in the -10 dBFS to -6 dBFS range just so I have some head room if someone decides to get excited.

I use to record the Cleveland Orchestra under George Szell. He was a master of levels and when he had the orchestra play a triple F you could set your console and tape recorder's volume at that level and he would never go over it.

I try not to add any extra reverb or eq unless it is a real problem and then I do it only in consultation with the performer or producer for the album.

I am finding that Classical artists are starting to worry about people listening to their material in their cars and want the overall levels brought up or at least the low level passages brought up to compensate for road noise which I personally find offensive since it changes the dynamics of the music in a way that the performer and the composer never imagined.

Best of luck in your mixing and mastering. Please let us know how it turned out.

[Manozi]

Hey guys,

Thank you for your replies - they're very helpful!

In fact, that's what my original instinct was - keeping the peaks somewhere in that area, but lately as I listened to the other guy's recordings I started to change my mind.

I'll keep them like they are, and maybe do some minor adjustments, but I'll stay clear of the normalization.


Cheers!
-Manol

[dave-G]

Originally Posted by Manozi

1) Is there a difference between pushing up all the faders a bit more in Pro Tools to get a better master level when doing the final mix, and applying 'Normalize' in Sound Forge to a completed mix (stereo track) to accomplish the same thing - bring the master levels up? (This is the software I'm working with so I'm using it in the examples) Would one sound different than the other?

Is someone mastering this record? If so, you can leave peak headroom as suggested, and let it be dealt with by the ME. Or ask him/her.

Otherwise, I'd say to use the headroom like the other engineer did, but within your Pro Tools session. If the whole of your contribution to the CD is in one PT session, this is easy; run it all the way through, with a meter on the master, or just command click on the fader's attenuation/boost amount level display to show the maximum peak level for the entire run. Then boost your master fader by say 0.3dB (or 0.5dB) less than the difference between that peak level and 0dBFS.

This should be more sonically advantageous than boosting via normalization on the resulting file, as the PT mix engine is 48bits, and moving the master fader is mathematically equivalent to moving all the individual track faders.

Good luck!
-dave

[bob katz]

Originally Posted by Thomas W. Bethe

I think that reynaud's post sums things up very nicely.

For classical I try and stay in the -6 dBFS to maybe a peak at -3 dBFS range and let the rest of the music lie where it is. I try never to use ANY compression or limiting or normalizing but try and do the recording in such a way that I don't have to play with the levels to much in post production and if I do I use the "gain" adjustment in WL for increasing the gain.

"Classical"? What is "Classical"? Symphony orchestra? Choir? Chamber group? Big crest factor or small? Each of these produces considerably different apparent level or loudness at the SAME POSITION OF THE VOLUME CONTROL.

The fact is that a small chamber group or soloist reproduced with a given peak level will sound often 10 dB or more too loud compared with a symphony orchestra with the same peak level!

So the answer to your question is complex. Experienced mastering engineers understand this dilemma and try to compensate for it, for example, leaving more peak headroom with a harpsichord solo recording than with Beethoven's ninth symphony. If I may put in a plug for my book, second edition, the new chapter 14 really tries to cover these and other dilemmas and make a Rosetta Stone suggestion for some consistent approaches to mastering levels in the 21st century.

[Thomas W. Bethe]

Originally Posted by bob katz

"Classical"? What is "Classical"? Symphony orchestra? Choir? Chamber group? Big crest factor or small? Each of these produces considerably different apparent level or loudness at the SAME POSITION OF THE VOLUME CONTROL.

The fact is that a small chamber group or soloist reproduced with a given peak level will sound often 10 dB or more too loud compared with a symphony orchestra with the same peak level!

So the answer to your question is complex. Experienced mastering engineers understand this dilemma and try to compensate for it, for example, leaving more peak headroom with a harpsichord solo recording than with Beethoven's ninth symphony. If I may put in a plug for my book, second edition, the new chapter 14 really tries to cover these and other dilemmas and make a Rosetta Stone suggestion for some consistent approaches to mastering levels in the 21st century.

You mean I have to purchase yet another copy of your book.....YEA! I can't wait to read it.

It is the one book I always can recommend to young wantabee mastering engineers to get them started. When is the 2nd edition due out?

[elpillo]

Originally Posted by Thomas W. Bethe

You mean I have to purchase yet another copy of your book.....YEA! I can't wait to read it.

It is the one book I always can recommend to young wantabee mastering engineers to get them started. When is the 2nd edition due out?

It's already out.

Amazon.com: Mastering Audio, Second Edition: The art and the science: Books: Bob Katz

[Manozi]

Originally Posted by bob katz

"Classical"? What is "Classical"? Symphony orchestra? Choir? Chamber group? Big crest factor or small? Each of these produces considerably different apparent level or loudness at the SAME POSITION OF THE VOLUME CONTROL.

The fact is that a small chamber group or soloist reproduced with a given peak level will sound often 10 dB or more too loud compared with a symphony orchestra with the same peak level!

So the answer to your question is complex. Experienced mastering engineers understand this dilemma and try to compensate for it, for example, leaving more peak headroom with a harpsichord solo recording than with Beethoven's ninth symphony. If I may put in a plug for my book, second edition, the new chapter 14 really tries to cover these and other dilemmas and make a Rosetta Stone suggestion for some consistent approaches to mastering levels in the 21st century.

Thanks for the additional insight Bob!
I appreciate the information.

It's always great to hear from someone with so much experience


-Manol

[SoundWeavers]

Manozi,

I'm no ME thus can't give advice on mastering.

But in recording classical style music, chamber, ensemble, quartet... I try to go for the room sound - have as little processing - instead work with mic/preamp selection and placement/positioning. I also try to understand the dynamics of the material by asking the performers to go through the quietest and loudest passages and adjust the gain accordingly. I will tend to set the gain to -6dbFS peaks knowing that it will most likely hit -3dbFS during actual performance. I tend to also put a limiter on the chain just in case the performers get really amped up.

with proper mic'ing and gain staging, you will have a recording that is easy to mix and easy to master.

FWIW, I've started hearing classical-style recordings that border on the square wave - such as a Vanessa Mae CD (which is pop I know, but still kinda styled as classical).

Bob's book is GREAT and I've had it for 6 months but I haven't gotten to Chapter 14 yet. :D the great thing about it? I can understand and appreciate most of what he's saying, too.

[Manozi]

Thanks for the reply SoundWeavers!

I am just listening thru the recordings now and my levels are in the same range - peaks at -6 - w/ the occasional one going up to -3Db to -2Db.

I think I will keep them as they are - I tried messing with levels in Sound Forge, tried normalization as well - and I hear artifacts in the sound - it messes with my mix. I'm happy with the sound itself and I wouldn't want to compromise that so I'll stick with what I have

Thanks to everyone for replying once again.

Cheers!
-Manol

[Don S]

A question I had when I first started. Coming from a classical performance background, I had specific ideas on how to aproach this. I agree with most engineers that noticable compression is not desirable. But where to put levels (during mastering) has to do with the number of players and the dynamic range. I try to keep the MF (mezzo-forte) dynamic somewhere between -15 and -10dbfs. I sometimes do use very gentle compression to control headroom. I'm a believer in "immitate before innovate". So, my decisions probably came from listening to well recorded performances and comparing them to my own.