Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[fader8]

Hi,

My first post here. Be gentle.

Anecdotal Background:
Over the past 24 months, I've had the pleasure of recording a concert pianist in her home. An interesting process actually. She'd practice a piece like mad for two months, then the tuner would come in (and bitch about not being allowed to move the mics), then I'd come in (and bitch about the tuner not stretch tuning the way I wanted) and we'd lay down a 2-track take, followed by finely crafted cocktails, mmmm.

So now we're done. I finally get my Neumanns and my best pres back. That's a good thing. I must say the results are stellar. The Schimmel Grand has a sweet sound that just lulls you. Anyway, I digress.

I tracked through RME converters into Sequoia. Just because I could, I recorded all the takes as 96kHz 32bit floating point files. Now, I'll do all my editing and audio housekeeping in Sequoia, but then I'll be taking the files and converting them to 24bit regular to move them into Logic Pro. Why? Because I need to squish this material into Redbook and my Mac has the UAD-1 Precision Mastering Suite which I'm presently enamoured with and is the best tool I have for this job. That, and the pow-r dithering when I go to 16bit.

The Question:
Should I maintain unity throughout the 32 to 24 conversion, or would there be any benefit to normalizing some of the quieter pieces before conversion?  It occurs to me that the conversation between the RME and Sequoia is only 24bit, which i guess in theory means that there isn't anything there below the 24bit bottom that can be recovered anyway.

Am I thinking straight on this, or is there some convoluted digital voodoo somewhere that I'm missing? Am I "safe" just doing a direct file conversion to 24bit? Maintaining 96k of course. Ya' know, Save As.....

Thanks in advance for any advice. Truly.

[Ronny]

Fader8 wrote on Tue, 31 January 2006 15:56

Hi, My first post here. Be gentle. Anecdotal Background: Over the past 24 months, I've had the pleasure of recording a concert pianist in her home. An interesting process actually. She'd practice a piece like mad for two months, then the tuner would come in (and bitch about not being allowed to move the mics), then I'd come in (and bitch about the tuner not stretch tuning the way I wanted) and we'd lay down a 2-track take, followed by finely crafted cocktails, mmmm. So now we're done. I finally get my Neumanns and my best pres back. That's a good thing. I must say the results are stellar. The Schimmel Grand has a sweet sound that just lulls you. Anyway, I digress. I tracked through RME converters into Sequoia. Just because I could, I recorded all the takes as 96kHz 32bit floating point files. Now, I'll do all my editing and audio housekeeping in Sequoia, but then I'll be taking the files and converting them to 24bit regular to move them into Logic Pro. Why? Because I need to squish this material into Redbook and my Mac has the UAD-1 Precision Mastering Suite which I'm presently enamoured with and is the best tool I have for this job. That, and the pow-r dithering when I go to 16bit. The Question: Should I maintain unity throughout the 32 to 24 conversion, or would there be any benefit to normalizing some of the quieter pieces before conversion?  It occurs to me that the conversation between the RME and Sequoia is only 24bit, which i guess in theory means that there isn't anything there below the 24bit bottom that can be recovered anyway. Am I thinking straight on this, or is there some convoluted digital voodoo somewhere that I'm missing? Am I "safe" just doing a direct file conversion to 24bit? Maintaining 96k of course. Ya' know, Save As..... Thanks in advance for any advice. Truly.

Depends on what type of normalization you are talking about. Is it peak normalization or RMS normalization. More AE's use normalization than ME's, IMHO. I don't use peak normalization because the noise floor goes up the same amount of dB's as the increase to peak. Not clear if you are mixing in Sequoia and than mastering in Logic or trasnfering multi-tracks to Logic, mixing and mastering there, but not using normalization very often I would transfer the multi-tracks or the two track mix to Logic without normalization. That way I can control the noise floor when I do any further processing. If the tracking is done well with low noise floor from the get go, and peaks are relatively high, you may be talking about negligible differences anyway and it's really hard to give finite advice without hearing the material.  

[fader8]

Thanks Ronny,
First, if I normalized at all, I was thinking peak, to -6 or maybe -10. My goal is to absolutely preserve the dynamic range attribute and not let anything get cut off the bottom during this conversion. I'd have to be more conservative with RMS normalization as it might imply some limiting of peaks and I don't want to disturb the crest factor at this part of the process. My question is more scientific, i.e. I don't expect to hear an audible difference at this point in just the floating to fixed point conversion. . I'm more worried about later when I'm using the UAD multiband on the Mac to expand some very quiet passages up. I want to be sure I'm not losing any music down there, even if it is close to, or even in,  the noise floor.

Second, the noise floor, electrically and acoustically, is quite acceptable, in fact negligible. I've got some pedal thumps, no big deal. There's no mixing, it's a 2-track recording. Refreshing, huh? Just convert and master.

I'm inclined to agree with you that no normalization is the way to go. I guess I'm just covering all the bases. Doing it right, first, will save me grief later.

Thanks again Ronnie.

[fader8]

Right, I looked over the quieter files, did some analysis. I'm golden. I won't lose a thing if I do a straight file conversion. Plenty of legroom there. I think I was over-worrying the issue. Must be a symptom of age. I'm blaming everything on that now.
Cheers.

[astroshack]

Frankly, leaving Sequoia for Logic just for the UAD-1 plugins is a compromise I would not make.
I use Sequoia...and I have 4 UAD-1 cards and own all the plugins. I also own most of the TC Powercore Firewire plugins, plus all the Algorithmix and Waves plugins. So I do know what all the plugins offer. But Sequoia was created by people obsessed with recording and mastering high quality classical music in Dresden. It is very capable of doing everything you could possibly need for this piano recording, without using any UAD plugins. Surely the idea is to do as little as possible to the original recordings?

I assume you own Sequoia and therefore have version 8.2?
If so, you have the wonderful built-in Analog Modelling Suite plugins designed by Sascha Eversmier (of digitalfishphones fame). These sound VERY good and should be very useful to you. If your recording is good, you will have very little need for EQ, but the linear phase EQ within Sequoia is fine for your needs. So is the excellent room simulation if you want to add ambience. There are some very good impulses included on the CD. Sequoia also offers very useful compression/limiting tools (in addition to the AM Suite dynamics tools), plus Pow-r dithering. You can burn your entire CD without ever leaving Sequoia and without ever changing the original files.

To me, it seems a no brainer: stick with Sequoia all the way.

Sean    

[astroshack]

Just one thing to add: if you downsample in Sequoia, use the highest quality sample rate conversion setting.
But for slightly better results, you may want to look at r8brain pro from Voxengo. The quality of the SRC you choose will make a fairly big difference to the eventual 44.1kHz files.

Sean

[fader8]

Sean,
I own Sequoia 6, which I believe is what was current at the time we started this project. I deliberately didn't change anything over the course, so it never got upgraded. Also, I'm not sure I could make use of it as the most powerful Win machine I have is that one, a Sony Vaio 500mHz, half gig, with a firewire drive. Leaned out for 2-4 track 24/96 recording and editing.

I won't be buying another Windows PC as I couldn't keep up with their maintenance and I was tired of buying a new one practically every 18 months. So upgrading to Sequoia 8 doesn't seem practical, as much as I like the software. I can appreciate your preference for many of it's fine features, but as you said, doing the minimal is the rule and for the very, very minimal stuff these recordings need, the UAD mastering plugs should be  more than adequate. Although, I'll let you know if my tune changes as I get further along!  

For the bulk of my usual work with sound design, post and music, Logic, Protools, Max/MSP, Final Cut Pro and particularly Kyma are my main arsenal. I need a lot of MIDI and Logic delivers that and has great audio routing features. My capital equipment budget is real tight now and when it is available I'm more inclined to buy more processors for the Capybara than buy a new PC and upgrade software that is used so little. I know where you're coming from. I'm just considering the practical aspects.

Good point on SRC. I'm going to try Peaks new algorithm but will move on if not satisfied. BTW, what's the "last word" in hardware based SRC these days? Might be wise to just rent something.

Thanks so much for your input.

[astroshack]

Oh, OK. Now I understand the logic for the swap to Logic.

Regarding software SRC, r8brain pro is the best of the cheaper options....and Saracon is the best, period.

you can see a current test of many SRC algorithms being run by some members of Glenn Meadow's mastering webboards:

http://src.infinitewave.ca

cheers,

Sean