Likely a dumb question here, but just to be clear, by “exporting the stems”, do you mean that you are exporting players as separate files and then importing those individual files into your DAW for further processing/mixing?
Yes
Maybe not the ideal solution, but perhaps you can just boost the output of your low tracks in the Dorico mixer before you export your stems. You can bring them back down when you mix in your DAW.
That’s what I am doing, regardless that I still need 24bit or 32bit audio output.
Manually pushing volumes up (then turn the stem volume down in DAW) may scramble the audio sample quality of stem files.
24 bit is useful in live recording situations since you have additional headroom - so if a musician plays unexpectedly loud you will not clip. But numerous blind tests have shown that no human being can detect the difference between 16 & 24 bit recording.
Also - you say your stems are peaking at -30? Are they going to peak at -30 in the final mix? If so, then there is no issue at all - there is no need to boost them in the first place. And if you need to boost them in the final mix - then just boost them in Dorico. Going from 16 to 24 will serve no purpose here.
If you’re curious, here’s a fairly good explanation: https://people.xiph.org/~xiphmont/demo/neil-young.html
It’s not about detecting the difference between 16 & 24 bit raw recordings, but about side effects of processing 16 vs 24 bit recordings.
I am familiar with the 1990s tricks for getting the most out of 16 bit audio files. I am simply requesting that we not have to use them in Dorico in 2020.
An excerpt from the link you shared:
An engineer also requires more than 16 bits during mixing and mastering. Modern work flows may involve literally thousands of effects and operations. The quantization noise and noise floor of a 16 bit sample may be undetectable during playback, but multiplying that noise by a few thousand times eventually becomes noticeable. 24 bits keeps the accumulated noise at a very low level. Once the music is ready to distribute, there’s no reason to keep more than 16 bits.
I’d like to add something after what pat_ has shared just before. I have been trained as a sound engineer in the late 90s, early 2000. I never heard the difference between 48kHz and 192kHz recordings, but I can tell you that the difference between 16 bits and 24 bits is sufficient enough, on classical music with a nice natural reverberation, for me to tell without errors which recording is broadcast in 16 bits and which one is 24 bits. So I’d say that once the music is ready to distribute, there’s no reason to keep more than 24 bits. Unless you don’t really care about detailed reverb (space). Rant over.
1 Like
Exactly that is what matters in mixing context, after importing some (hopefully 24 bit) stems to your mixing project, and later in mastering. It is different to process (and deteriorate) a 16 bit wave than to process (and deteriorate) a 24 bit wave and then at the end to convert it to 16 bit. Both final 16 bit waves will have different quality.
Paul Walmsley, are you saying that prior to the render Dorico is internally using 32 bit? IE: for any inserts?
AFAIK all “standard” VST instruments and effects use 32 bit floating point for data transfer within the VST host. There are even 64-bit VSTS and VST hosts available for those who think it makes a difference.
Individual VSTs could use lower bit depths internally, but I would be surprised if any of the mainstream sample players (Halion, Kontakt, etc) did that by default.
Of course a VST might intentionally use a lower bit depth - for example 8-bit audio is a genre in its own right.
Note, the “32 bit” label is technically accurate but might be misleading. The bit depth of 32 bit floating point is actually 23, but there is always 23 bits of accuracy, for any practical setting of the gain controls, even if the signal level is way above or below 0 dB. The total dynamic range of 32 bit floating point audio without loss of quality is something like 1500dB!
Yes, it’s my understanding that the internal signal path is 32-bit float.
To restate: we will add the option to export in different formats in the future - just for the moment there are some architectural issues that make it non-trivial to add a dialog devoted to audio export options.
2 Likes
[quote=PaulWalmsley post_id=868382 time=1560189177 user_id=71556
The real utility of a 24-bit output is if you are exporting stems, which may be at a lower level, or if the final output is due to be further processed (eg mastering). If, however, you are producing a final wav/mp3 output directly from Dorico then IMHO I don’t think you’d get any real benefit over 16-bit.
[/quote]
Thanks Paul - that’s mostly what I need and you have the reasons exactly right, at least from my limited corner of the world. If a music supervisor or library owner wants 24 bit stems or alts and the final “master” is being done with dialog etc. to picture. They are quite happy with mp3’s for submission/approval/review. My technical criteria on this issue is mostly - do I get paid. 
Metadata is kind of a thing since we’re talking about exports, but it doesn’t kill me to do that in another tool.
Regarding iTunes Plus file submission, Apple enforces a requirement of 48000Hz 24bit AIFF file to either iTunes Store or Apple Music.
Anyway, I guess Arne Wallander is gonna have a great talk with Daniel regarding the vital importance of providing 24bit audio output.
To my experience, NotePerformer is good enough to be used for final audio production.
(I dub live-performed string first chair tracks with NotePerformer string ensembles, and it sounds great.)
(Stay safe, people. Health is the most important overall.)
1 Like