This exact issue has been the bane of my existance- especially with the "lower end" clients... It is a major consideration, though. My recording of the Berlioz Requiem I did last June in Sydney had a dynamic range of over 50dB. Even after mastering, there was roughly 40 dB, but the choir director liked it so I let it be...

I do a couple things. The first thing I do to deal with dynamics is that I will manually change it. I insert a linear cross-fade in Sequoia on the audio. I then extend the fade so that it is several seconds long. At that point, I can raise or lower level and nobody will notice. I'll region off a loud section and raise everything around it. I have just raised the overall level without doing a thing that can harm the transients or perceived dynamics of a piece. If you do this right, you can easily remove 10-15 dB of dynamic range from a recording with complete transparency.

Parallel compression can also help you with this... You add dynmics and a bit of body to the sound but you manage to maintain your transient information on the uncompressed side. Result again is an RMS level change but not much change to the sound of the music...

In the end, it is somewhat just the issue of dealing with classical/acoustic music. It won't be as loud as the rock. If I had a dime for every director that complained that the recording distorted when it got loud (and then I find out he's listening on a car stereo and he has it turned all the way up to hear the quiet music), I'd be a rich man...

--Ben

As I stated in my first post, i am very much interested in the dynamic treatment of classical material. I have to deal with this issue on a weekly basis since we broadcast our concerts on the venezuelan tv every saturday night. The way I approach this is "some of the dynamics but not all of them". A typical example is the beginning of Ravel`s Bolero. Every percusionist makes a point of honor on how soft or "pianisissimo" can play that snare drum part. Well , I must say I manually have had to bring that part up until I get a reading of -12dbVU (being -1dbVU the loudest level of the recording). When the first solo comes in (I think is the flute) I try to establish terraces of level adding 2db and manage to keep the dynamics within those terraces untouched. This applies for the Bolero which is a constant crescendo piece. When I reach the final part, which is pretty loud(about 3 minutes from the end) I try first to remove any peaks that make no significant contribution to overall volume and last but not least, I will use a moderate amount of compression, with readings of -2 to -4dbVU of GR.
I try to avoid the squashing effect by any means since that´s a no-no in classical, even for tv. I try to respect the dynamics "within" the different episodes or phrases, the trade off in order to obtain a level that will keep the viewer from touching the volume control is that the F (forte), FF (fortissimo), and FFF(fortisissimo), will all modulate at -1dbVU (short transient peaks removed)and that P(piano), PP(pianissimo) and PPP(pianisissimo) will modulate at -12dbVU. That should "raise the floor" enough to appreciate the softest passages.
Nedless to say that ambient noise at concert hall becomes quite critical. I hope this is of any use. I hope even more that somebody could give me some advice since this procedure is very time consuming.
Best regards to all.
Javier

I'm relieved to hear that I'm not alone in this.

I have been reticent to even bring this up in some classical circles, as I know some 'audio nazis' that would think less of anyone who 'ruined' a piece by inflicting any kind of adjustment to anything in it. These are usually the same folks that think it possible to capture the 'true sound' of a piece. "Just record what's there". Strangely, there are some pros who believe that they do this! One engineer I know has ONE pair of mics and ONE recorder. He beileves that this setup records what his ears really hear, and never changes his rig. I myself like his tapes sometimes, but often as not something is wrong that might have been fixed with a different pair or a spot...

While many music directors want dynamic problems solved, they also would freak if they knew how many adjustments occur to the so-called 'natural sound' in the course of making the CD work to their satisfaction. Of course, most experienced folks know that to get a listenable CD a lot goes on behind the scenes-- but I don't think many of them really have any idea what it takes. And they don't WANT to know- fine by me, as I am happy to be trusted.

When I am contracted to record a piece, I always get a score and a few examples of previous recordings of the work. At the library, these CDs might have been recorded any time in the last 25 years. Many of them suck IMO. Often the dynamic is so extreme that the quiet parts are almost unlistenable. The recording often only reaches within a few DB of full scale once in the entire album-- say on the loudest field drum smack. I often wonder how much/ little care actually went into these albums!

Well, that's my current rant!

I myself love parallell compression and maybe a little hand limiting on the loudest transients. For me the final test is to put the CD on and listen while doing someting else and paying the music minimal attention. I take notes on 'funny spots' and go back to correct them. Often, the strange area I need to fix is a TOTALLY unaffected area that sounds unnatural even though it is absolutely pristine. Then I fix it!

more thoughts?

regards

Exsultavit

Mmmmmmm....Dynamic Range (Drooling in my best Homer Simpson voice...)

There are several different schools of thought on classical music's use of extreme dynamic range. I've got a huge CD/LP collection at home and it's very interesting (and quite easy too) to hear the differences in what levels of dynamic ranges studios employ.

Personally, my philosophy has always been "Mix for the best system on the market." In other words - don't dumb down the recording for the yutz who is driving around in a 68 Impala with Pioneer 6x9's shoved in his back windshield. I'm assuming that the guy listening and buying my music has a stack of Mark Levinson Amplifiers, B&W Nautalis speakers and an acoustically sound room. Mainly because I don't want these people to think my recording sucks.

So, in other words, I go for as much dynamic range as humanly possible. (As long as I'm not exaggerating the natural dynamic range of the orchestra - exceptions noted.)

First, I like to make sure the room is registering <-60dB noise. It's hard to find many concert halls much better than this. They do exist, but unfortunately Telarc has the contracts with these orchestras already. :( (I have made decent recordings in rooms with noise measuring -45dB before. I've also made numerous recordings within a particular hall around this area with a -30 dB noise. Notice, I didn't say those were good recordings. BTW, all of this level gathering is done during sound check. I ask the conductor if possible to choose portions of his music that represent the loudest and softest passages for the sound check. Bear in mind, during the performance, the level will be at least 3dB higher.)

I then will find the loudest, most climactic portion of the piece of music and ensure that this part is at approx. -3dB. (This is usually done in post production) There will always be transients that are significantly louder than this, so I employ a limiter with a threshold of -1dB with a max output level of -.2 dB. This ensures that there are no "overs." If I notice that my meters are constantly hitting the -.02 mark, I quickly realize that the piece has been mixed too loud.

Once the overall levels have been set, only minute adjustments should be made. This will preserve the dynamic range in the recording. Most of the recordings that I've done that I would classify as "good" are those with at least 55dB of dynamic range. (Not including reverb tails - only meaning from loudest note to softest note at the points of articulation or peak level)

J...

Cucco-

Now THAT's dynamic! I myself have a different ideal-- but that's obvious.

As far as overs-

I always do a live 2trk mix recorded to the client CD, a DAT, and 2 tracks of the multitrack. I'll also record (to another 2 tracks of the HD) a version of my live mix that is 10db down from the "hot mix". Between these, I am pretty safe from overs on my mix. I also record all mics to their own tracks. The individual tracks are where I usually go (obviously) if there is mixing in post.

sincerely

Exsultavit

<>

Jeremy, you're gonna LOVE Sequoia. That's one of the tools/tricks I use for final mastering, etc. There's a very useful, transparent brick-wall limiter available on the output for just this sort of thing. Overages are pretty much eliminated this way. You can also lower the entire project in 6 db increments, esp if things are out of hand and you need the extra headroom during a mix.

Sorry, I seem to have hijacked the thread over to Samp/Sequoia again....

As for limiting, I'm reading a lot of similar approaches to what I deal with here, as well. (Enjoyed Javier's Bolero description, and it makes sense, esp in regards to making a broadcast.)

I try to think ahead on where the project will end up, and not unlike Jeremy, I first make a 24bit "Safety" master that gets saved "untouched". This is my 'Audiophile" safety copy, before anything else gets done to it, and I save it. (My thinking is that someday perhaps they'll want to do a DVD-A with it, etc.)

Then, depending on where the project is ultimately headed, I have to make some choices.

Client CDs sometimes need a little manipulation, esp if they're going out in mass quantities. (And as everyone noted already: people often don't understand why their choral recording of Messiah doesn't kick ass the way Zitney Spear's latest does, on the boombox.) Radio & TV broadcasts are another consideration, because it WILL get squashed horribly if you don't take care of it yourself. (Again, sometimes I will even save & label a "broadcast" version.)

NB: I'm not talking about the hideous "Finalizer" crap that gets done to rock/pop recordings in some BS excuse that "It's only going to get squashed by the radio people anyway." WRONG!
I"m talking about making it fit in the general confines of a typical minimally compressed Classical or NPR station.

You will also get some funny looks from the station engineers if you try to slip in raw things that drive their Optimods crazy. Many radio stations have "dead air" alarms that will go off after a specified amount of time, and it gets embarassing if they keep calling to ask you why your CD's are getting them into trouble with the FCC. For broadcasts, I still try to keep applause levels as my peaks, and work down from there, occasionally bringing up the quieter passages in a way that isn't heard as pumping or maniupulating. (With Samplitude's object based editing you can do this seamlessly, but there I go again...) .

In the old days, with "soft" analog tape compression and much more forgiving overages, you could do announcers at -10 or -12, and let fly the big bombastic music, with occasional really high peaks and excursions into +4, +8 and more. In the old days, Broadcast FM radio was ok, there was often enough tube-based compression (and optical stuff) to do it smoothly. THESE days, if you give 'em a CD or DAT with an announcer that low, chances are they'll crank it up to 0, and then all kinds of terrible things begin to happen to your music (which could now be peaking at 10-15 db ABOVE this).

I have a client that can't tell the difference in ambience & timbre between two wildly different halls edited into the same track, but he goes NUTS when he hears ANY kind of compression on his typmani and other wide dynamic passages on the radio. (So much that he's threatened to pull his shows off the air....) Talk about your sonic idiot savants! (Does anyone ELSE notice this with some classical "purists?")

Anyway for some really wild dynamic range on certain things, I learned to work around this by making the announcer and the applause the HOTTEST thing on the CD, while the music only peaks at about the same place as the announcer's normal speaking peaks. By older standards, it's ass-backwards and completely wrong, but in today's world of "NOTHING OVER ZERO" digital peaks, it's working out nicely for most broadcasts we're doing, and we "hit" the Optimod a whole lot less during the musical passages. Granted, some people do still have to reach for the volume control in quiet passages (and get their ears blown off when the big parts come back in), but it sure has reduced my client's & listeners' complaints about compression and level changes. I'm fortunate too in that this is for an NPR station that doesn't use much compression at all for Classical, and a little faster version in the evening, when they switch over to Jazz.

But we could talk for days about how to gently, unobtrusively "Tame" things that needlessly rob dynamic range, etc. For the most part, I start out with NO changes whatsoever, wherever possible, and stay pure with it. But as we all know, the playback world is too noisy and too distracted to live with even a 50 db dynamic range, so we often have to make a few nips and tucks, here and there. Mainly, I feel if you can "hear" it, you've gone too far.

Just want to chime in that this is one of my main issues. Bigs ear (well, eyes), over here listening for your experience and learning. he "real slow" volume change sounds like a good thing to learn. (I do use Samplitude, here we go again).

Gunnar Hellquist

JoeH wrote:
" I first make a 24bit "Safety" master that gets saved "untouched". This is my 'Audiophile" safety copy, before anything else gets done to it, and I save it. Then, depending on where the project is ultimately headed, I have to make some choices."

That exact same thing I do. Since we record audio direct to a Sony dsr11 DVCAM VCR, which is set to record at 48Khz/16bit, I make sure that we don`t go into overload (peaks around -3dbVU) and then I take that audio from the original avi file and save it to a data cd. As and audio and video coordinator, it is one of my duties to keep the Orchestra`s archives, so everything goes untouched to "the vault".

JoeH wrote:
"In the old days, Broadcast FM radio was ok, there was often enough tube-based compression (and optical stuff) to do it smoothly. THESE days, if you give 'em a CD or DAT with an announcer that low, chances are they'll crank it up to 0, and then all kinds of terrible things begin to happen to your music (which could now be peaking at 10-15 db ABOVE this)."

Some of the guys here at the TV station`s master control room go
crazy if they don´t see the Betacam SP´s VU meters almost touching the red. So when I transfer to that analog machine, I make sure that nothing goes over Odb, which is the threshold level they have for their limiters to kick in very heavy, unpleasant limiting. And then I post a frame on the video tape warning them about the softest audio levels of -15db and the peaks at 0db. In other words, I tell them to stick (no matter what), to the reference tone.
I had to do that after one broadcast of Richard Strauss "Zarathustra" in which they rose the level of the trumpets on the very beginning to almost 0db. You can imagine what happened a few measures later when the orchestra hit that triple Forte.
I prefer our listeners-viewers at home to rise their volume once (about 1/3 of the scale up) so they do the make up gain with the power available on their receivers. Anyways they know by now that our broadcastings have a softer audio level.

Best regards to all. Javier