Summing in Cubase: when to change resolution?
pietrogirardi
Thread Starter
#1
27th December 2009
Old 27th December 2009
  #1
Gear nut
 

Thread Starter
Thumbs up Summing in Cubase: when to change resolution?

Hi everybody!
The project I've been working on is almost finished. All ITB, all with virtual instruments and loops.
All projects in cubase (SX3) are set to 32 bit float, 44.1Khz.

It's summing time, before delivering tracks to the mastering engineer. I wish to send him tracks at 32 bit float, 96 Khz, because AFAIK all plugins benefit from a higher res (especially reverb - I work with 96 khz IR's).

Question: is it right to
- simply export audio, and in the export window select "32 bit float, 96 Khz", or
- FIRST I must set the project to 96khz (with all the annoying issues related to .wav audio files conversion and time shifting) and THEN export 32/96?

Thanks,
Peter
#2
27th December 2009
Old 27th December 2009
  #2
Lives for gear
 
Geert van den Berg's Avatar
 

I'd check with your mastering engineer first if his software can handle 32 bit float files.

24 bit files are much more common.

Quote:
Originally Posted by pietrogirardi View Post
All projects in cubase (SX3) are set to 32 bit float, 44.1Khz.

It's summing time, before delivering tracks to the mastering engineer. I wish to send him tracks at 32 bit float, 96 Khz, because AFAIK all plugins benefit from a higher res (especially reverb - I work with 96 khz IR's).
Says who?

I would not bother with upconverting to a higher res if your recordings have been done at 44.1kHz. You should have done the recordings at 96kHz.
LQM
#3
27th December 2009
Old 27th December 2009
  #3
LQM
Lives for gear
 
LQM's Avatar
 

Best to check in advance as stated, but it shouldn't be a problem, people with a lot of VSTs or freeze tracks tend to export in 32 bit float quite commonly these days, although mixing to 24/96 should be equally fine.

You may also find that Voxengo's R8Brain does a better job of SRC than Cubase's internal processing - most DAWS don't have the greatest SRC capability. The free version of R8Brain will typically do a better job of changing the SRC. Or if you have the latest Soundforge (10 which uses the Izotope SRC) and/or Izotope Ozone, the Izotope SRC algorithm is also highly regarded.

I agree though that it's typically better that unless all of your tracks are virtual instruments and effects (in MIDI and not rendered out as audio) that the benefits of upsampling something recorded at 44.1 Khz are debatable at best.
pietrogirardi
Thread Starter
#4
27th December 2009
Old 27th December 2009
  #4
Gear nut
 

Thread Starter
Quote:
Originally Posted by Geert van den Berg View Post
I'd check with your mastering engineer first if his software can handle 32 bit float files.
Already checked, he can handle 32 bit files.

Quote:
Originally Posted by Geert van den Berg View Post
Says who?

I would not bother with upconverting to a higher res if your recordings have been done at 44.1kHz. You should have done the recordings at 96kHz.
As said before, I use virtual instruments. With virtuals you do not record anything - MIDI is the only language spoken until summing: in this final passage, the virtual instruments played via MIDI are effectively translated into audio files.

The difference between the two choices are:
1) if I change the project settings and then sum, and I have some audio tracks together with MIDI tracks (sometimes it occurs, even if audio tracks are not more than 2 for each project), those audio tracks are upsampled and often moved from the original position on the grid.
So, summing this way needs an additional job in adjusting those audio tracks. But this way I'm SURE that everything works in 96k.

2) Second choice: setting 32/96 only in the summing (cubase="export") window just gets rid of these problems - no upsampling / coversion of the single audio tracks in the mix, but just a 96khz stereo file output, regardless of the project settings.

My doubt: these two ways of summing bring to the exact same result?

By instinct I would say "no" because in the second way, Cubase should just take the 44.1 project, read it in 44.1, plugins work in 44.1 (for example, IR1 with 96k IR's downsample to 44.1 while processing) and then just create an upsampled 96khz stereo file: that would be useless to me because, during mixdown, I want the sound of VSTi's to be processed at 96khz by plugins (for ex., the reverb).

Did anyone have this doubt, and solved it someway?

Thanks,
P
pietrogirardi
Thread Starter
#5
29th December 2009
Old 29th December 2009
  #5
Gear nut
 

Thread Starter
....maybe I should reformulate?
#6
30th December 2009
Old 30th December 2009
  #6
Lives for gear
 
nuthinupmysleeve's Avatar
 

Personally, I would bet money that nobody could tell the difference between recording your virtual instruments at 96k vs. 44.1k in a blind test. Seriously... don't kill yourself with a poor workflow for no reason, IMHO.
#7
30th December 2009
Old 30th December 2009
  #7
Lives for gear
 
Geert van den Berg's Avatar
 

Quote:
Originally Posted by pietrogirardi View Post
....maybe I should reformulate?
No you're perfectly clear. I read over the Virtual instrument part in your first post.

I think then the best option is to upsample before doing the bounce.

You might want to downsample that bounce and compare it with a bounce which has not been upsampled before the bounce. Compare and report back...

But like psalad said, I also think you will only find a marginal difference.
#8
30th December 2009
Old 30th December 2009
  #8
Lives for gear
 
dotl's Avatar
 

...
#9
30th December 2009
Old 30th December 2009
  #9
Gear addict
 
MikeMitchell's Avatar
 

Quote:
Originally Posted by psalad View Post
Personally, I would bet money that nobody could tell the difference between recording your virtual instruments at 96k vs. 44.1k in a blind test. Seriously... don't kill yourself with a poor workflow for no reason, IMHO.
I could not agree LESS. Not all virtual instruments sounds significantly different at 96k but a few have a completely differently tonality. It's based on the coding I imagine. I recall one song in particular I used a virtual minimoog (I've had a few, I believe it was the minimonsta) for the bass line, lead, and a few pads. I swapped the project from 48 to 96 and it was drastically different, and thats not in a way that people say they hear a night and day difference b\w converters. This was massive. All triggering from midi, no recorded audio.

No scientific testing, just my real world experience.

I'd say roll at 96 if your pc can handle it and it sounds audibly better to you. I wouldn't necessarily say some of my vi's sounder better at 96k, they just sound very different. I find if I develop a patch around whatever rate I'm at it sounds best at that rate, for the most part.

-m
#10
30th December 2009
Old 30th December 2009
  #10
Quote:
Originally Posted by pietrogirardi View Post
Hi everybody!
The project I've been working on is almost finished. All ITB, all with virtual instruments and loops.
All projects in cubase (SX3) are set to 32 bit float, 44.1Khz.

It's summing time, before delivering tracks to the mastering engineer. I wish to send him tracks at 32 bit float, 96 Khz, because AFAIK all plugins benefit from a higher res (especially reverb - I work with 96 khz IR's).

Question: is it right to
- simply export audio, and in the export window select "32 bit float, 96 Khz", or
- FIRST I must set the project to 96khz (with all the annoying issues related to .wav audio files conversion and time shifting) and THEN export 32/96?

Thanks,
Peter
Hi Peter.
Your mix will not benefit from SRC unless you carry out the SRC before exporting, as the method of simply exporting a 44.1kHz mix to a 96kHz file will carry out the SRC as almost the last process, merely padding the file with zeroes.
When upsampling, it makes little difference what you use really. The sole exception could well be to use Min Phase mode in Voxengo's R8Brain Pro, but that will be a sod to do for a whole project.

Okay, if the mix is being mastered at 96kHz then you could argue that the increased SR will make for less quantization distortion (easier to get good sound via 96K DAC than 44.1k DAC) but that is about all - there will be no extended frequency response (all files recorded at 44.1) and no cumulative benefits from the 96K multitrack, as you mixed at 44.1

Not at all sure it's really worthwhile unless you use top flight soft or hardware SRC and carefully tweak all the parameters to get the best though.
Unless you can convert the entire project (save a copy to a new folder first) and render at 96K then all you are really doing is padding the files
pietrogirardi
Thread Starter
#11
30th December 2009
Old 30th December 2009
  #11
Gear nut
 

Thread Starter
Quote:
Originally Posted by MikeMitchell View Post
I'd say roll at 96 if your pc can handle it and it sounds audibly better to you. I wouldn't necessarily say some of my vi's sounder better at 96k, they just sound very different. I find if I develop a patch around whatever rate I'm at it sounds best at that rate, for the most part.

-m
I do agree. Also, I've noticed a sensible difference in summing at 44.1 and at 96 - the latter sounds exactly like the mix itself, while a 44.1 mixdown sounds narrower and less 3D... Don't know if it depends on the host or the plugs.

Quote:
Originally Posted by neilwilkes View Post
Hi Peter.
Your mix will not benefit from SRC unless you carry out the SRC before exporting, as the method of simply exporting a 44.1kHz mix to a 96kHz file will carry out the SRC as almost the last process, merely padding the file with zeroes.
When upsampling, it makes little difference what you use really. The sole exception could well be to use Min Phase mode in Voxengo's R8Brain Pro, but that will be a sod to do for a whole project.
This was exactly my doubt. You confirmed me what I feared - exporting a 44.1 project to a 96k stereo file just pads the file with zeroes.
The solution is to convert the projects (upsampling the - very few - audio tracks via R8Brain, or letting Cubase do the job) and then mixdown at 96k.

Thanks slutz!! I'll let you know when the job is done....
Peter
#12
30th December 2009
Old 30th December 2009
  #12
Lives for gear
 
Flying_Dutchman's Avatar
 

you work in 44.1, ok
so, export 44.1
get a good dithering tool at stereo out, voxengo limiter or whatever is good
export in project sample rate and export 16bit
24 bit , as said, is common cause protools can only do 24 bit, but here is no technical reason, if you got to export to mastering, phone the guys
if they can do 32, do 32 44.1, if they can do 24bit 44.1, do this
i see no advantage in exporting a 44.1 project to 98k master, did i miss something?
#13
31st December 2009
Old 31st December 2009
  #13
Lives for gear
 
nuthinupmysleeve's Avatar
 

Quote:
Originally Posted by MikeMitchell View Post
I could not agree LESS. Not all virtual instruments sounds significantly different at 96k but a few have a completely differently tonality.
OK, that's all well and good, but that is a big IF... IF it sounds better of course do it. I have heard no difference in any of the softsynths I use. I have done enough listening tests and have read up enough on the science behind sample rates that I'm confident there is no real world difference for most content.

That said, use whatever you'd like. If you *think* you hear enough of a difference, and that's good enough for you, then that's good enough. For me, my take is there is no difference unless it has been tested... as I read (and believe) there is far more to human perception then hearing a difference in the moment.
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
fn86 / Rap + Hip Hop engineering & production
0
enigmacrypt / Music Computers
0
lclyman / Gearslutz Secondhand Gear Classifieds
8

Forum Jump
 
Register FAQ Search Today's Posts Mark Forums Read

SEO by vBSEO ©2011, Crawlability, Inc.