I wasn't able to find much information on this for classical (and soundtrack and everything else for that matter).

I would just send them the final mix as an uncompressed 32 bit float file
and leave the final decisions about peak level and compression/limiting to them. Leaving them more room to adjust allows them more options to achieve the best end result.

I try to leave between 6 and 18 dB free on the top for them to play with. I won't compress the 2-buss at all or the main pair but I might occasionally compress spot mikes to help hide them. LUFS will vary all over the place because that's just how it goes when you have a lot of dynamics.
--scott

if i'm not mastering myself, i don't aim anywhere other than not getting any clipping or distortion during production.

if i'm mastering myself (for cd/dvd release), i hardly ever go higher than -18dB lufs for classical music; peak (assuming there is no isp) is mostly -0.3dBfs.

mastering for broadcasting is another topic, as well as for playback on pa systems or for backing tracks - and i'm not yet even talking about streaming or vinyl...

short: times for a single master that does it all imo are long gone!

Avg LUFS at -18dB, reserve peaks at -2 or -3dB if going for external mastering, peaks at -1dB for my mastering efforts,

Jnorman. How do you get -1 peaks without using a compressor? Unless you mean for the actual master, right?

Not for anything recorded?

For capture, I never like seeing the absolute loudest thing on the recording go over -6 and I keep my averages at around -18.

D.

Doug - no, recording peaks are kept at -12dB or so. I use “purestgain” plugin from airwindows on the master buss to adjust LUFs. Though I do often use pro-c for very mild compression at times (it has a pretty smooth preset called “tighten classical music”), but I don’t use compression to add gain.

I think you should ask the mastering engineer for her or his preferences. Some of them post their desired specs on their website, but if not just send the engineer an email and ask. It'll save you both some time.

The mastering engineer I worked with on our most recent project simply told me to keep true peaks below -3dBFS; no guidance at all on LUFS.

A lot of studios have this information right beside the service description.
I would look at the webpage or just ask the engineer directly.

This is usually what I aim for.

That 'go higher than' part is important. Because on a harpsichord piece you might be able to watch the meter bouncing up and down a little bit over the course of the movement, but with a full romantic orchestra you might see a huge, huge difference between the quiet bits and the loud bits.

Whether you want to accurately reproduce and maintain that difference depends a lot on what your expected listener base is. My inclination is to always go for accuracy of reproduction but I'm not the producer, I'm just the engineer.
--scott

See image....

As always, YMMV.
All the best,
Mark

my reasons are less fancy/sophisticated: -18dB lufs is just about what some european radio stations still allow and also where i often land without much processing on the mix bus IF i can implement my preferred 'modern way' of recording and mixing (which is largely based on spot mics).

the producers i get to work with here are less technical than musically savvy so as long as i can provide evidence that i'm within an acceptable 'range' in terms of the overall soundstage, i'm mostly given decision-making authority, certainly on technical details.

___


now 'accuracy' imo is a more interesting topic - in terms of what?

i give very much about it when it comes to localisation of sources within the stereo or surround soundfield and select main mics accordingly although there is repertoire that imo requires a much more blurry image...

if accuracy relates to dynamic range, i tend to disagree that unaltered dynamic range is a desireable thing anymore...

[...which is kinda ironic as we today have the means to capture, mix and distribute music with insane dynamic range.]

accuracy in terms of room impression remains critial although i don't care whether to capture it (in excellent and quiet rooms without any audience in) or to fake it; more often than not, i find the latter to yield more pleasing.

accuracy in terms of spectral balance imo remains to be the last frontier: far too many people have a pretty good umderstanding of how unamplified instruments sound so i tend not to mess with filters on mains and ambis much (if at all) while on spots, imo it's often necessary, as well as dynamic and efx processing.

___


anyway: up to -18dB lufs for classical to my ears seems to be about right, up to -12dB lufs for rock/blues as well (and jazz/fusion somewhere in between) while anything beyond or actually above that, i need yet to hear a mix that i would find very pleasing...

[some folks who are in to pop/metal etc. are laughing at me for not wanting to pump up mixes to 'competitive levels'!]

Sounds good. But if it's higher, you can always reduce it in post. I'd certainly expect more processing for delivery to radio than for delivery to the mastering folks for distribution on CD.

I think most people would agree with you, in part because of the way the music is ultimately used by the end-user. Few people sit down and carefully listen to a symphony recording all the way through these days. But I like to.
--scott

so do i - but i assume that both you and i have slightly better than average home hifi (or listen in our studio)?

___


[i love my tad/augpurger/kinoshita's! - too bad they are in a studio in
which i don't work that often anymore (although i co-own it: it's just too far away to commute)]

If the mastering engineer is getting 32bit float, why the anxiety over levels? If they are getting 24 bit fixed, then anything up to -1 or -2 will be fine? Never understood why this was ever a topic worthy of discussion.

I write and produce my own cues for television, for network forensic and crime news shows, I’ll aim for -14 lufs , but often will sit just behind that at - 16 ish.

Like dee, for classical type writing on dramatic scripted shows, where I’m featuring just piano and strings, or pads, and wrapping my cues around dialog, I’ll go with a range of -18 to -23 lufs. The entire show is going to be mastered, so I don’t worry too much about loudness, but try to stay where the best fidelity lives.

For both scenarios, I’ll set the threshold anywhere from - 0.3 to - 1.0., I’m more concerned with crest factors than anything else, so I’ll often use volume automation rather than outright limiting to get an even crest factor - the last thing you want is for a music editor to have to ride your gain too much.

These numbers sound really good to me. I use ozone 9, fab filter and some of the pro tools pro series plugins.

Indeed - leave the mastering to the Mastering Engineer (?!)

Or is this about one doing one's own mastering?

The point here is that there is no one magic number that works for all recordings. It depends on the content. As the engineer, you need to find that sweet spot where the music wants to live. You can't force something that is outrageously dynamic into the same level as a John Luther Adams piece that is all steady state tones. People get all caught up in numbers, but they are a guide to make material fit into a continuum of material out in the world. The experienced engineer knows when the record is at the right level. If you are doing broadcast, then sometimes you need to leave a little average or peak level on the table in order to make something sound good. At the end of the day, these are maximums, and to make good sounding recordings you need to know where to make the trade offs.
Unfortunately, this is where experience and taste comes into play. I tell clients all the time that their record has reached it's loudness potential. Any louder and it is to the detriment of the music.
As always, YMMV.
All the best,
Mark