mastering for multiple formats -- workflow in DP

[schmargle]

I'm working on a few projects that don't have any budget for mastering, so as the mixing engineer, I'm preparing the songs for CDs and digital downloads. I'm using Digital Performer to prepare all the fades, get the levels right, do the spacing, compress, EQ, limit, etc.

The original tracks for one project were recorded at 24/48. In the mastering project, I've kept the project at 24/48 because I want to preserve the highest quality audio for creating AAC files and MP3 files (and also keep a "mastered" version at 24/48).

Currently, I bounce the whole project, which has each of the songs starting at its own marker. I then take the full-album bounce and use Peak to slice up the songs. I send the songs to iTunes to create the AAC and MP3 files.

The catch is that I also need to produce 16-bit/44.1k files as well.

If I bounce the whole thing at 24/48 then covert to 16/44.1 within DP (by using the covert audio file command in the soundbite bin), the peaks come UP a bit--by about .4 dB or so. That's not a huge deal, but my goal is to keep the peaks to -.2 dB for both the digital file version and the CD-file version. If I bounce the whole thing at 16/44.1, the same thing seems to happen. If I do the conversion in Peak, it's the same.

So, somehow, converting from 48k to 44.1 after the whole thing is bounced (whichever way I do it) adds a few dB of something.

To get around this, I've set up two mastering sequences within the same project--one for 48k and one for 44.1. I've had to duplicate the fades and spacing for all the songs (not a huge deal, but kind of a pain). I've converted the soundbites BEFORE the processing (making new 44.1 versions of the files) and then bounced a CD-file version separate from the other version. This time the levels for both versions turn out as they should--at -.2 dB.

But what a pain! ...And of course, if after listening to this for a while, the band wants to change a mix, I have to reimport and place the file in both sequences...and bounce the whole dang thing again two times.

So what is a better workflow for all this? If you had 24/48 files and you needed to produce...
16/44.1
24/48
AAC (320 kbps) (has to be done in iTunes or Logic)
MP3 (192 or 256 kbps)

Is there a way to do this more efficiently within DP? Without having multiple sequences and multiple bounces?

I suspect this might be easier in Logic, but I like using the DP MasterWorks EQ in the signal chain (and I don't want to pay to port those to Logic). It might also be easier in Pro Tools, but again, I want to use MW EQ, and I also don't feel like listening to the whole darn thing bounce in real time each time a re-bounce it.

Any suggestions?

[martian]

not sure why you getting that rise.

I've probably had similar but I just didn't kill myself over the .2dB diff.. especially since most times it was going thru a dolby encoder later. which would have line ups to do anyways

maybe what you can do to save a bit of time is cheat DP.

just make sure the bounce for each track is exactly the same length. save the times in the control panel! recall those times for each bounce

now you can just replace the audio files in finder for this session you have created.

so when you open up this "different dither" session - DP will recognise the files have changed and update the waveforms..

obviously you have to match the file names exactly..

sorry I cant help on the minor level issue

[davedempsey]

I always do my SRC in Peak. I always see 0.1db added to the 'peak' when converting 24/48 to 24/44.1 - expect that to happen always and prepare for it at the previous stage. When translating 48000 approximations per second into 44100 approximations per second you have to expect some alterations in the result

[Armageddon]

I generally only master out of a DP project in the Project Format (88.2/32-bit float for CD projects, 96 kHz/24-bit for movie work that will ultimately be delivered at 48 kHz/16-bit) with -.3 dB headroom. I then load the two-track master into Wave Editor and do all my conversions into a new file. After DC Offset, SRC and dithering, I generally gain about -.1 dB or so, which I end up correcting by normalizing to -.3 dB again.

[schmargle]

Thanks for the replies. Given your responses and the other stuff I've read, I think I'm going to bounce in 24/48 and do the sample rate conversion after. I guess the trick will be to lower the volume by like at least .4 dB before doing the SRC. If I don't, I wind up with a 16/44.1 file that has a peak at 0...which makes me think that there is some little bit of clipping somewhere.

Thanks again.

[mikehalloran]

>AAC (320 kbps) (has to be done in iTunes or Logic)<

No. It can also be done in QuickTime 7.6.6.

Install QT 7.6.6 in the Utilities folder (it must be that version). It is downloadable from Apple and is found as an extra with the OS 10.6/7/8 installers. Do not move it to Applications or it will conflict with QT 10 and neither will work properly.

If you have an old license for QT 6 or 7 Pro, you can use it to access the additional functionality in 7.6.6.

If you have QT 7.7, trash it unless you are on OS 10.5.x.

[bolla]

If I bounce the whole thing at 24/48 then covert to 16/44.1 within DP (by using the covert audio file command in the soundbite bin), the peaks come UP a bit--by about .4 dB or so. That's not a huge deal, but my goal is to keep the peaks to -.2 dB for both the digital file version and the CD-file version. If I bounce the whole thing at 16/44.1, the same thing seems to happen. If I do the conversion in Peak, it's the same.

I checked and I'm getting similar results. Must be due to how DP interprets the inter-sample value.
I couldn't find a way within DP to get the maximum peak of a track without having to play the loudest sections. I ended up using Wave Editor from Audiofile. Is there a way in DP??
Cheers Bolla.

[Killahurts]

I checked and I'm getting similar results. Must be due to how DP interprets the inter-sample value.

Could be.. Or, DP might be placing a dither on SRC. That could explain the small discrepancies.

[Dan Worley]

Yeah, I don't know why but the SRCs I use (MAS or Core Audio) always add a tenth of a dB for some reason. I'm not smart enough to know why that is (interpolation maybe?). I also don't know if there are some SRCs out there that don't do that.

The MAS SRC is adding four tenths of a dB to your levels? That doesn't seem right. Try exporting from the Soundbite window and see if Core Audio adds that much with its SRC.

Though I don't always follow it, I think it's good practice to keep the peak levels down away from digital zero, just for situations like this, and for other reasons as well--like playback through cheap DACs. There's just no way you're going to hear a difference in loudness with peak levels down around -0.5 dBFS to -1.0 dBFS, instead of right at FS zero. I've read some papers that suggested the peak levels should be down around -3 dBFS to ensure best performance on typical playback devices and systems.

[David Polich]

A mentor friend of mine once said, "the difference between 44.1 and 48 isn't even a lousy octave".

I think you ought to convert everything to 24/44.1khz. The quality difference at 24/48 is basically negligible. Mix your tunes, then do a real-time bounce to stereo track and export that as 24/44.1khz AIFF out to Peak. You say you have Peak, I don't know why you're not just doing all the conversions you need to do in Peak. You can make multiple copies of each master in Peak, each a different version - MP3, WAV, AIFF, FLAC, whatever.

Since it doesn't look like anything is going to video/film, nothing is going to end up at 48khz, you might as well convert things
down to 44.1khz and stay there throughout. It saves you a sample rate conversion in the end.

As an aside, let's say you have your mix peaking at 0 db on your master
fader. When you export it to Peak, you'll see the peaks (no pun intended)
come up at around -4.00 average. This is why I don't master in DP.

[martian]

+1 - that sounds like a solution!

I'm going to admit that for what I do 24 bit 96K is a waste of disk space... I'm dubious about 24 bit in many applications.

maybe for a flute recorded in a really great quiet room... but with the worlds hums and buzzes and whispers - I dont think many can hear it... maybe its cos I come from the hardrives cost a fortune era...

[Shooshie]

24/88.2 here. The separation of instruments, the depth of field, and the detail are incredible. I can tell the difference when I bounce down to 16/44.1, but it's better, IMO, than bouncing from 24/48 to 16/44.1. I think some of the separation is preserved.

I'm probably full of crap, as there is nothing scientific about my choices. But I CAN tell the difference when I record in the 24/88.2 format. I'm doing classical music, and I know what I'm trying to get. I can't get it when recording in 16/44.2 (and believe me, I've tried). I do get it with 24/88.2. The people I record are aging professionals who have played everywhere, for everyone, in some of the best organizations in the world. They've heard professional recordings of their works for all their lives. They get excited when they hear my recordings of them, because they've never heard it like that before. It's more than just the format, of course, but that's a big part of it. I just record it the way I'd like to hear it. Thank goodness for technology, MOTU, and DP.

Shooshie

[schmargle]

>AAC (320 kbps) (has to be done in iTunes or Logic)<

No. It can also be done in QuickTime 7.6.6.

Hm. I didn’t know that. It must be same engine for all those things, eh?

I think you ought to convert everything to 24/44.1khz. The quality difference at 24/48 is basically negligible. Mix your tunes, then do a real-time bounce to stereo track and export that as 24/44.1khz AIFF out to Peak....Since it doesn't look like anything is going to video/film, nothing is going to end up at 48khz, you might as well convert things down to 44.1khz and stay there throughout. It saves you a sample rate conversion in the end.

Yeah, for this music (only semi-dynamic rock), I agree, I can only barely hear the difference between 44.1k and 48k. But I want to keep a mastered version at 24/48 for converting to the compressed formats (MP3, AAC, etc.). Even if we're dealing with very small differences, it seems like you should start with the highest quality audio before dropping it into the sausage grinder. And I also want to have a final, mastered version at 24/48 just in case some day CDs go away and we actually start to listen to things at a better fidelity.

I have another project that we recorded at 24/96, so whatever I learn here will be applied to that project as well. And I can definitely hear the difference between 24/96 and 24/44.1.

You say you have Peak, I don't know why you're not just doing all the conversions you need to do in Peak. You can make multiple copies of each master in Peak, each a different version - MP3, WAV, AIFF, FLAC, whatever.

Ultimately, I probably will do the conversions in Peak. But actually, the same thing happens in Peak as DP--there's a little gain added when you do SRC. I'm just striving for streamlining the process and avoiding the need to remove or add gain to compensate for SRC.

I guess what I’d really like is something similar to the workflow in Logic, where you can bounce in multiple formats at once and actually choose the sample rate of the output (which, I think you used to be able to do in DP but can’t any more). Of course, the issue there is that you’d have to bounce each song separately within the mastering session--if you don’t, you’d have a full-album MP3, for example. And then if you bring it into Peak to cut it up, you’d have to re-convert it again, which would be bad.

Bouncing each song separately from within the mastering session would be okay, except I don’t trust that the transition from one track to another would be seamless. With something like Peak, I feel more confident that there wouldn’t be a hiccup when one tune ends and the next begins.

As an aside, let's say you have your mix peaking at 0 db on your master
fader. When you export it to Peak, you'll see the peaks (no pun intended)
come up at around -4.00 average. This is why I don't master in DP.

Hm, that's weird. If I bounce something in DP through a limiter set with an output of, say, -.2 dB, that's how it shows up for me in Peak. I've never had any discrepancy there. Not sure what would be happening for you.

24/88.2 here. The separation of instruments, the depth of field, and the detail are incredible. I can tell the difference when I bounce down to 16/44.1, but it's better, IMO, than bouncing from 24/48 to 16/44.1. I think some of the separation is preserved.....I CAN tell the difference when I record in the 24/88.2 format.

Yeah, I totally agree. Once you get out past 16/44.1 into 24/48, 24/96, and beyond you can totally hear differences, especially with classical, jazz, and even a lot of rock. Human hearing might be limited to about 44.1 (20 Hz to 20kHz), but there’s definitely a difference when we capture a larger frequency spectrum, and certainly also (in the case of bit depth) a greater dynamic range.

[schmargle]

Quick update: I think I'm going to have to do a separate DP project for the 44.1k mastering. No matter what bit of software I try, doing the sample rate conversion creates little spikes that are not so little (is that what aliasing is?). So basically, even pulling down the total gain by 1 or 2 full dB before doing the SRC isn't enough to avoid clipping. So, I need to do the SRC before running it through the compressor/limiter.

As I do more research, I've been finding mastering engineers who basically bring mixes into the mastering session/project (A-to-D) at 44.1k so they can avoid that in-the-box SRC. What a pain. Maybe in the future I'll look into more dedicated mastering tools, like Ozone.

If you're interested, one resource that I've found is helpful in comparing SRC of various applications: http://src.infinitewave.ca/ DP7 is not bad compared with some others, but it's not perfect.