Limiting and compression in Classical orchestral releases

[klaukholm]

So the loudness wars finally reached our operation. We recorded a first class string ensemble with percussion.

Much of the material lives in the piano and softer dynamic range, but the peaks have a massive sound with gran cassa punctuations of the peaks.

We left the material uncompressed as usual with the peak at -3 dB.
The composer remarked that the cd volume was overall very low.

We raised the volume 6dB by limiting the master bus and the peaks are now at -0.2dB.
effectively we have reduced the dynamic range of the recording with 3db.

The peaks are still perceived as softer than on a commercial pop cd.

I am curious to know how my colleagues approach this issue.

[jdjustice]

i'm certainly no ME but this sounds outrageous! limiting on a classical release?
i know it has been done occasionally (VERY LIGHTLY)......
.....but the whole idea sounds ridiculous to me and in bad taste.


~j.d.

[The Listener]

I am surprised about that question. We have volume knobs - we turn them up when listening to wonderfully recorded classical music - fff should be extremely loud, ppp barely audible and that's it... Compressed, limited and otherwise "pop" treated classical music (sometimes for comercial usage) sounds like sh*t!

If it doesn't compare to other classical recordings then it must be something wrong with capture itself - not good enough microphones, not the right colour, not the right balance, etc.

If it is only much quieter than pop-rock recordings and also some jazz - that's normal...

I ocasionally listen to the art music radio and it is much more quiet than the rest of radio programmes, but it has a beautiful sound.

So, compare commercial classical recordings to your classical recordings and don't compress !!! HIgh quality EQ should do the trick of bringing more details, more "punch" if needed...

We - the classical music listeners are used to that. We understand why it is not as loud as pop CDs, we actually love the dynamics! Some of us have audiophile systems that do justice to such uncompressed and wonderfully recorded music with full frequency range (no lo shelving...) and extensive dynamic range.

Educate the composer, :)

[klaukholm]

We have not encountered this kind of request previously and would personally prefer to use the volume knob rather than loosing dynamic range.
Ultimately the client is requesting something and we want to make him happy.
A long string of contracts depend on this client being satisfied with the final product.

He is actually a very accomplished composer that we enjoyed working with.

The piece starts out with ppp non-pitched bowing (white noise)
after that two string players enter with artificial harmonics, the piece grows steadily and ends in a massive grancassa hit.

[d_fu]

Originally Posted by klaukholm

The piece starts out with ppp non-pitched bowing (white noise)
after that two string players enter with artificial harmonics, the piece grows steadily and ends in a massive grancassa hit.

I think it is legitimate to compress lightly. Not every listener will be able to experience the full dynamic range of such a recording in his surroundings and with his sound system. The crescendo effect is not lost with a light compression.

IMHO it is not wise to have the music play at e.g. -30 dBFS or so most of the time, just because a few loud impulses reach full scale. I would not want a listener of such a recording to have to run for the volume knob in panic, just because such passages suddenly become much too loud. This IMHO would disturb the listening experience much more than a slightly reduced dynamic range,which is hardly noticeable if well-done and does not necessarily have to destroy the effect of the dynamics of a piece.

Trying to preserve the "original" dynamcis (by what standards and from which position, the listener's or the microphones'?) by all means seems a bit ultra-orthodox to me.

Daniel

[d_fu]

Originally Posted by The Listener

Compressed, limited and otherwise "pop" treated classical music (sometimes for comercial usage) sounds like sh*t!

Not necessarily...

So, compare commercial classical recordings to your classical recordings and don't compress !!!

I'm quite sure that a lot of commercial recordings are lightly compressed or even limited at extreme peaks - but if it's well done, it won't be noticeable.

I ocasionally listen to the art music radio and it is much more quiet than the rest of radio programmes, but it has a beautiful sound.

And it certainly uses compression...

HIgh quality EQ should do the trick of bringing more details, more "punch" if needed...

Now that's what I would call "pop treatment"... I practically never use EQ...

[Jim vanBergen]

I used to believe that classical music used no compression...

It's a misnomer...compression has always been used in various forms, either simply in the mastering stage, in some form of tape compression, as well as in the radio broadcasts who are looking for a dynamic range of no more than 15db!

I encounter this problem constantly. I have raised my +4 dBU analog = -14dbfs level up to -0.5 dbfs by use of an analog "tape emulation" gain stage, "in order to allow our client to perceive the maximum CD loudness as the same as other CD's". Conservative use of compression, in terms of either gentle limiting or dual path compression (one path compressed, one path uncompressed, both paths mixed together for the final product) have worked well for my clients.

For me it's more about having the same maximum volume and trying to achieve as close a dynamic range as you can while maintaining a realistic replication.

I know some purists will really piss on this concept, and I understand their position...but I only do what my clients want, based upon THEIR wants & desires.

[Legacy]

Hi Kjetil,

I've encountered this as well mostly with contemporary classical pieces that can have a dynamic range that is only realistic for a sound-proofed room. In these rare cases I prefer to use "De-Expansion" or 'bottoms up compression' where below a certain threshold the low level details are brought up but the louder passages are left alone (or lightly limited). This way the loud impacts are still as exciting or dramatic.

Or as Jim vanBergen mentioned - parallel compression is worth a try.

Pyramix's Solera does this very nicely as does the TC Sytem 6000. Anyway I suspect you are familiar with the idea already.

Hope all is well,
Silas

[The Listener]

Hi again,

d-fu, sorry for me sounding maybe too orthodox before. I am not totally against compression in classical music - like you said - unnoticable taming of the peaks sounds like a sensible think to do... And yes, I am aware that also art radio uses some broadcast processing including compression and limiting, but much less than the "pop" radio.

I am not fond of recordings sounding agressive like some cheap "remasters" of 60s classical recordings or some h'woodesque soundtracks...

I just happen to listen to Gorecki's 3rd Symphony with Dawn Upshaw and it is mostly around -30 and -20dfs... in silent parts... But the peaks are at ca. -0,2 - so it is using the whole digital range...

I don't know... if it is ppp it should stay ppp... I guess.

I am simply saying - don't compare classical to pop production - bringing up "white noise bowing" and artificial harmonics could sound really unpleasant and strange when boosted... Compare your recording to Decca, Deutche Grammophone, classical stuff on ECM, etc.

A subtle parallel compression comes to mind. And if the picture is blurry - masterful useage of EQ - which is not "pop treatment" when done as a sort of artful sound sculpturing with bringing the life sound picture alive in mind... If the microphones, preamps and positioning didn't do the trick already.

And just that - people that jump for remote when the "fff" hits don't know how to listen to music. Contemporary classical is not a harmless background soundtrack - it should be listened to attentively. And when you do that you don't kill the loud parts - they overwhelm you - just my thought.

Sorry for the long post.

EDIT: sorry, I checked CK recording site and you guys look very serious... I am even more surprised by the question. There is no way that a classical recording will sound as loud as a pop one. Or you'll really destroy the sound. The composer should now that, I guess. Don't bring the loudness wars to classical scene, please. I like loud when it should be loud, not all the time. Imagine -10df RMS string quartets - killing experience!

[teleric]

Originally Posted by d_fu

I think it is legitimate to compress lightly. Not every listener will be able to experience the full dynamic range of such a recording in his surroundings and with his sound system. The crescendo effect is not lost with a light compression.

IMHO it is not wise to have the music play at e.g. -30 dBFS or so most of the time, just because a few loud impulses reach full scale. I would not want a listener of such a recording to have to run for the volume knob in panic, just because such passages suddenly become much too loud. This IMHO would disturb the listening experience much more than a slightly reduced dynamic range,which is hardly noticeable if well-done and does not necessarily have to destroy the effect of the dynamics of a piece.

Trying to preserve the "original" dynamcis (by what standards and from which position, the listener's or the microphones'?) by all means seems a bit ultra-orthodox to me.

Daniel

I totally agree 300% with everything you said. We apparently of the same school. I'm curious as to know who were your teachers.

Eric

[Yannick]

When the time comes the composer doesn't like his own dynamics ...

One thing that helps is taming the loud hits during tracking, by the musicians themselves. A good hit really together sounds more powerful than just banging the hell out of the instruments, and it will leave more headroom.

If we're correcting after the fact, well we're doing just that aren't we ?

Sometimes contemporary music seems to have too big a dynamic range, but I fail to see the point in reducing this. It will never be easy listening music, or something to listen to in the car.
Unless we're actually correcting a balance problem.

Reminds me of the soprano that wanted to listen to her own recording at realistic levels, on a run-of-the-mill hifi - and then complained it distorted.

[matyas]

As Silas alluded, contemporary music sometimes demands a slightly different recording aesthetic than standard rep classical. (I'm saying this as a composer as well as an engineer.) Sometimes those huge dynamic ranges that you see in some contemporary music are as much a timbral effect as a dynamic effect, so a little compression/limiting in the recording process (for purposes of clarity) isn't necessarily a bad thing. Another point to consider (and it doesn't sound like this is relevant in the case of the piece being discussed), many composers these days are as influenced be jazz, rock, or techno as they are by standard-rep classical. Some composers want their recordings to sound like a rock record, rather than an audiophile release.
On a tangent, but sort of apropos Jim's point, I was listening the other night to the old 1964 Harmonia Mundi recording of Boulez's Le Marteau sans Maitre on vinyl the other night. Tape compression, vinyl compression and EQ, tubes and transformers in the recording path, probably some limiting at the mastering stage - all the stuff rock guys love and classical guys tend to avoid these days. But it still sounds great - incredibly detailed and lifelike, in fact. So maybe a bit of limiting and color isn't necessarily a bad thing!

[The Listener]

I agree with you Matyas on many things, but if you would put on a contemporary pop LP (they still exist), it would be much louder at the same level setting. Now, getting that kind of loudness is something I hope noone tryes to achieve.

[fifthcircle]

I deal with this on a daily basis... I'll give a ref to a client and then I get a call that goes like this:

"Hey, what's up with this CD? It is distorting all over the place"
Me: "Huh? Explain further- as you were sitting here in my studio, we did not hear any distortion"
Them: "Well, I put it in my car stereo and I had to turn it way up. Then it got loud and it was distorting all over the place"


Or you can substitute car stereo for boom box, home theater system, etc... As a matter of course, to avoid this, I now incorporate some dynamic manipulation in all of my recordings. More often than not, it is a manual process. I'll region off the very loud sections (using linear crossfades of varying length) and raise the soft sections so that they are a bit more audible. More often than not, this works just fine. I may also do anywhere from 1-4 dB of peak limiting for the loudest sections.

The extreme case of this is an album (a piano trio) I mixed and mastered a few months back. Long story short- the average rms for the album was -12. I have rock albums that were softer than this. When they asked for it louder, I just said "no- it is too loud already" I used parallel compression, limiting and every other trick I could think of. In the end, I asked for the released CD to not include my name.

The things we do to satisfy clueless clients.

-Ben

[Gerax]

I'm working 50% of my time with classical material, most of the time recorder with some of the finest players in Italian orchestras. I'm just finishing the "mixes" for a classical CD project I recorded early on this year, which includes a full size 52 piece symphonic orchestra with solists, and I must admit that coming from a pop background...well I like it loud, but being educated to a healty amount of classical music I try to respect dynamics and tone as much as I can. This doesn't mean that I cannot take advantage of the full dynamic range afforded by the digital medium. I admit that I almost always use a limiter (more than a compressor) on my master fader, even if only for taming some of the highest spikes that would force me to turn the whole level down by a few dBs to avoid digital clipping. If the limiter is set right the gain reduction effect is unnoticeable (almost...to engineers ears, let alone common listeners in their less than perfect environments), so I do use it. With the final medium being 16/44.1 CDs I think that if the program recorded goes below a certain level (like in ppp sections) then it's going to be covered by dither noise (if the material has been recorded at 24 bits and brought down to 16 for CD), so what's the point in not raising the level those few dBs while still in 24 bits and try to preserve al the possible detail and information? Limiting can be harsh if overdone or done in an improper way, but using a transparent limiter and the right settings won't harm the music, it will enhance it (IMHO), just like the right EQ touch by those few dBs in the right places.
I've never been an integralist that belives "you can't do that in classical music", to me there's no right or wrong way to to it, there's just a way to have a good and natural sound, and if that takes more than 2 or 3 mics or EQ and limiting to achieve...so be it. Comparing POP CDs to classical CDs loudness-wise is almost pointless: frequency range and average dynamics are completely different so I just think one working with classical material should apply gain reduction just enough to contain the digital clipping and to raise the average RMS by those few dBs that would make the CD listenable in most situations without wondering if the CD player is working all the time or not . A bit conservative if you want, or delicate, but definitely not a "no no".

Cheers

L.G.

[pkautzsch]

A little narrowing down the dynamic range isn't a bad thing, I think.
The main concern is: do you hear the limiter working? If yes, it's not good. And most times, you *will* hear it on some gran cassa hits or other peaks.

Fader riding respectively volume automation is the key IMHO.

And, for contemporary music, I agree that there are different, and not that strict, rules.

[Yannick]

On the other hand, our studio is calibrated to the K20 standard.

It rarely happens we need to raise the overall level/limit/etc to adhere to that standard. Most extreme contemporary material fits in the K20 standard. Just because an acoustic instrument also has its limits.

So if it fits this standard - why lift the quiet parts or limit ? To end up with K17 or K14 ? Because pop is going towards K10 or 8 ?

[dsomeister]

I believe you could try parallel compression, as a possible answer. The Sonalksis
compressors works well on an effects track in a DAW. Ratio around 2:1, threshold at
-40, may have to use a good eq. before the compressor to take out a bit of the
grand casa. The trick is mixing the original with the Parallel compressed signal. I have found that around a -12 level of the PC signal works well with the Uncompressed tracks. We routinely use a Metric Halo 2882+dsp to feed cdr master a parallel compressed signal .

Cheers,
Roy Cherryhomes
www.soundboardrecording.com
Dallas Symphony Orchestra

[heathen]

I don't see anything wrong about using a touch of gentle opto compression on the highest peaks, there are some remarkably clean opto's around which would do the job beautifully, actually sometimes I feel enhancing spaciousness and ambience by keeping it gently in place, or some low level compression on multed stereo mix. Though perfect mic placement would be the key while tracking, working out who is quiet and who is loud in which piece and udjusting mic placement accordingly, Also riding the faders on mixdown would be essential through some passages.

[klaukholm]

Thanks for everyones thoughts!

It is useful to hear what everyone else is up to. It is good to see most of you are on the same page as us.

[larsfarm]

Wouldn't it be so that acoustic music is often intended to be heard at some distance in a concert hall, church or other large room. The typical listener to live music is further away from the sound source than the mics. The dynamic range at the listener will be smaller than the dynamic range closer to the sound source where the mics are. So couldn't that result in an artificially large dynamic range in the recording?

L

[d_fu]

Originally Posted by larsfarm

Wouldn't it be so that acoustic music is often intended to be heard at some distance in a concert hall, church or other large room. The typical listener to live music is further away from the sound source than the mics. The dynamic range at the listener will be smaller than the dynamic range closer to the sound source where the mics are. So couldn't that result in an artificially large dynamic range in the recording?

Lars, I was considering to post the exact same thoughts... You beat me to it.
Very good points. Has anyone measured "real world" dynamic ranges at listeners' positions?

[pkautzsch]

Originally Posted by larsfarm

Wouldn't it be so that acoustic music is often intended to be heard at some distance in a concert hall, church or other large room. The typical listener to live music is further away from the sound source than the mics.

Right. See below for details.

Originally Posted by larsfarm

The dynamic range at the listener will be smaller than the dynamic range closer to the sound source where the mics are.

Why? The dBspl level will be smaller. Not the dynamic range.

Originally Posted by larsfarm

So couldn't that result in an artificially large dynamic range in the recording?

I don't think so.

The point about mics being closer than audience, or about more than two mics in contrast to just two ears of a listener, is that people in a concert audience do not only hear with their ears, but also with their eyes. They see who's playing, and therefore they hear soft violin tremolos which would on a recording be completely masked by brass. So we need to make the violins louder to make them heard on the home stereo where nobody can see them play.
Same goes for overall levels: a symphony orchestra can produce somewhat like 120 dBspl. You don't want 120 dBspl in your living room, and your neighbours don't want 120 dBspl in your living room either.

It's not about if, but about how and how much we should tame the dynamic range. I'd prefer not to hear the gear, but to hear the music.

[sonare]

Ben alluded to this earlier-- no one exists on a desesrt island-- not our clients, and not the people they hope will buy their CDs. I have had success in the past with the "it isn't loud enough in my car" type of client to simply ask, "have you compared it with the same genre material from D-G, EMI, Sony, etc etc?"

Usually I don't hear any more about it. I can only assume they do not realize we hang out on weboards and discuss such non-musical behavior as what they are asking for!

This would certainly be an appropriate subject for the upcoming Chamber Music America workshop in NYC in January-- what NOT to ask your engineer/producer and WHY.

They wouldn't consider something equally outrageous such as playing a concert in their pajamas, would they? (And I am not talking about a composer who thinks that is cool.)

Rich

PS-- 1 to 1.17 compression with judicious peak limiting will usually "tame" the dynamic range without doing serious aesthetic damage, IMO. That, plus a monitoring standard such as the "magic 83dB SPL."

[klaukholm]

Originally Posted by pkautzsch

Same goes for overall levels: a symphony orchestra can produce somewhat like 120 dBspl. You don't want 120 dBspl in your living room, and your neighbours don't want 120 dBspl in your living room either.

As a matter of fact we are measuring spl levels in the orchestra this month.
In the reverberant field we measured 60-108db in Verdi requiem which is a very moderate piece volume vice ,unlike the complete shostakovich we just played
I will come back with measurements as we progress. unfortunately we are not measuring extreme pieces like Mahler and Shostakovich. Sustained 120dB is not unheard of in certain positions in the orchestra. We are ultimately a fairly moderate orchestra volume vise