32 bit bounce

[marc-1]

Sorry if I missed something, (maybe there was an eaerlier thread not included) but exactly how does one do a 32-bt bounce in DP ? I see where you can export an audio file out of the soundbites window as a core audio export to 32-bit SD-II, but it follows that you have to have the bounce done before you do that right ? In which case, hasn't the permanent effects of dithering (or not dithering) to 24-bit already been printed ? What am I missing ?

[chrispick]

Sorry if I missed something, (maybe there was an eaerlier thread not included) but exactly how does one do a 32-bt bounce in DP ? I see where you can export an audio file out of the soundbites window as a core audio export to 32-bit SD-II, but it follows that you have to have the bounce done before you do that right ? In which case, hasn't the permanent effects of dithering (or not dithering) to 24-bit already been printed ? What am I missing ?

Yeah, you have to use the Core Audio SDII option. Doing so with the bounce window brings up a subwindow allowing you to choose 32-bit. That way, presumably, no downconverting from 32-bit back to 24-bit during bounce.

And yeah, Kubi, they pop up in Peak as 32-bit interleaved files.

[David Polich]

I'm still feeling dense about this. If you recorded at 24bit/44.1khz, then don't you have 24-bit audio files all the way through your project? And if so, what difference does a 32-bit bounce make?

I thought there was some heated discussion months ago on this forum about recording your entire mix in real-time as opposed to doing a BTD. The real-time recording was supposed to be much better than BTD. What happened to that argument? If BTD is so horrid, why would MOTU incorporate it into DP? Wouldn't the supposed sonic degradation be really really noticeable? I've just done straight 24-bit BTD's for like, forever. They sound fine to me. Why isn't 32-bit including as a project bit-depth option?

BTW, I gthink the advice from MOTU on dithering was, when bouncing 24-bit within DP (to create another track from several tracks, for example) you turn dither on. When doing a 24-bit bounce that won't be used in a project track, you turn dither off

I think it's hard to avoid losing bits here and there on the way to the 16-bit 44.1khz CD.

And with something as badly mixed and mastered as Christina Aguillera's "Stripped" album, who cares if a few bits get lost along the way?...

[chrispick]

I'm still feeling dense about this. If you recorded at 24bit/44.1khz, then don't you have 24-bit audio files all the way through your project? And if so, what difference does a 32-bit bounce make?

Yeah, your recorded files are at 24-bit, but when they're processed they're up-rezed internally withing DP to 32-bit for calculations, then down-rezed back to 24-bit. So, they're calculated at a high level, then rounded off, causing a truncation of the sound.

It's kind of analogous to what occurs when you posterize a bitmap image. It loses resolution in the subtle areas

This is why it's usually advised you have dithering on in DP. It helps this rounded-off data sound less truncated.

The idea behind bouncing at 32-bit is to prevent any up-rez-then-down-rez at the final mix stage. Then, mastering can be committed to the least-truncated audio file.

Now, when you down-rez for CD (16-bit), you're chopping away data again. But, you've at least made sure that all major processes prior to down-converting haven't resulted in an up-rez-down-rez rollercoaster (wherein, perhaps, a perceivable amount of data has been mucked with). And, within Peak at least, you can apply pow-R dithering (three varieties to choose from) that's said to be some of the best dithering algorithms available. Bob Katz digs it anyway.

I thought there was some heated discussion months ago on this forum about recording your entire mix in real-time as opposed to doing a BTD. The real-time recording was supposed to be much better than BTD. What happened to that argument? If BTD is so horrid, why would MOTU incorporate it into DP? Wouldn't the supposed sonic degradation be really really noticeable? I've just done straight 24-bit BTD's for like, forever. They sound fine to me. Why isn't 32-bit including as a project bit-depth option?

Well, I'm not one who believes that OTB mixing is better than ITB.

And yeah, I've been mixing and bouncing 24-bit for a while now and it sounds good to me too. I'm just trying an experiment to see if I can improve on that. Maybe any improvement will be too miniscule to register. I don't know. It's just a simple test for me to do, so why not?

Finally, I don't know why 24-bit isn't included as a project bit depth option. My thought on it is: When audio gear manufacturers introduced 24-bit tools, they thought it was the highest manageable bit rate given the existing processor and storage options. Since then, procs and throughput has increased in speed, affordability, etc.

Heck, I don't know.

[archtoper]

Well, I'm not one who believes that OTB mixing is better than ITB.

And yeah, I've been mixing and bouncing 24-bit for a while now and it sounds good to me too. I'm just trying an experiment to see if I can improve on that. Maybe any improvement will be too miniscule to register. I don't know. It's just a simple test for me to do, so why not?

Finally, I don't know why 24-bit isn't included as a project bit depth option. My thought on it is: When audio gear manufacturers introduced 24-bit tools, they thought it was the highest manageable bit rate given the existing processor and storage options. Since then, procs and throughput has increased in speed, affordability, etc.

Heck, I don't know.

I think that unless you have tens of thousands to spend on analog gear (and that could be in cables alone), it's best to say ITB all the way. For those that prefer OTB, I have to wonder if they did 32 bit bounces for their A/B comparison. If not, I'd feel the test was unfair as the all digital domain was not given it's best foot forward. I think when the MOTU boxes offer 32 bit DA is when we'll see internal 32 bit file formats (up from 24), but in the meantime the 32 bit bounce really sounds better. Actually, I think anyone not doing it this way is missing the boat. I've only had DP/PEak and a Mac for a few months but have been into recording for over 10 years (started on DA88's before there were discussion boards like this- heck- didn't even have email back then), and I've gotta say- these new tools are absolutely INCREDIBLE!!!

[David Polich]

Okay, I tried a 32-bit export and then a 24-bit export of the same file - and dang, it does sound better. Unfortunately, Peak LE won't open a 32-bit file, and DSP Quattro will, but when I then try to save the file in DSPQ as a 16-bit file (with dithering turned on), DSPQ hangs. Oh well - maybe I'll try this in Soundtrack Pro...

And yes, checking "dither" when doing a 24-bit bounce within DP does give better results.

[kelldammit]

well, let's use an example...albeit a grossly oversimplified one.
you've recorded a file. it's stored on the hard disk...nothing you do short of a destructive edit is going to change that file (the beauty of daws!).
let's assume for the sake of oversimplification, that that file's audio has a value of 1. we'll also say for the sake of argument, that you're only toggling bit number 24. so there it is, a 1 stored on the disk at bit 24.
once you hit "play" and that 1 hits the mix engine in DP, it becomes 1x10(1st power). now bump that slider up 1 dB. it's now 1x10(1st) x 1.4444444444444444 (not really the 1db log, but bear with me for the sake of illustration).
at this point, if you bounce to 24bit without dithering, your new 1.4444444444444 value becomes...1. the decimal places are completely lost (truncated) due to the 24 bit format not being able to represent the values below 1, other than 0.
so, you can dither back to 24 bit...in which your algorithm will attempt to round using some pseudorandom noise or somesuch...so your end result could be a 1, or a 2. hmmmm. you're erring almost as far either way (see why i picked .44444?). no matter what, the 24 bit format (in this illustrative case, anyway) simply cannot accurately represent values between 0 and 1.
if you export at 32bit, at least SOME of the decimal places can be represented due to the much lower values (levels) 32 bit can represent, so you may get 1.4444 out of the deal this way.
if you export 32float you get the 1.44444444x10(whatever power) that the mixer is sending (sort of, you might actually get something like 1444x10(-4th) or somesuch. float will scale the decimals back so that as much of the signal as possible can be represented in the 24 bits of the file, so in this case, the 24 bit "window" of resolution would have to be moved DOWN beause it's such a quiet signal. the other 8 bits represent the x10(x power) stuff.
so, the original file is almost meaningless once it hits the mixer...as the audio it contains is no longer truly represented as 24 bits from this point on. the 24 bit file is just a starting point for all the numerical madness that follows in the mix engine
another way to look at it: remember that files are only storage. you can open a file, and edit the heck out of it in a program (the program operates on a copy of the file in RAM) till it no longer even roughly resembles what you started with, but no changes to the file itself are actually made until you save (using that original file name).

hopefully that was SLIGHTLY clearer than mud?? and kind of sort of accurate...if not, i stand to be corrected

kell

[David Polich]

Hmmm - my eyes kind of glazed over halfway through that.

The only thing that matters to me is whether it sounds good to begin with, and if it sounds as good or better at the end. Whatever the bit depth actually is is irrelevant to me.Heck, if it was an awesome sounding 15-bit file that's fine.

It doesn't stop at whether it's a 32-bit bounce. Once you start mastering you begin bringing in all the variables associated with that - including how insanely loud you have to make everything, destroying dynamics along the way, adding distortion, etc. And of course, it's supposed to be 16-bit 44.1 at the end. Does it end up really being 16-bit, or is it 13 or 14 bits?
Does it matter if the end user is listening on a cheap boombox or stock car stereo? I love the No Doubt records, but you can clearly hear distortion and clipping on those CD's.

When you're doing your own mixing and mastering, you can get carried away with all the attention to details. As Brian Eno once said, nobody ever finishes a record, they just give up on it. At some point you have to give up on it and declare it over, and you might be surprised at how good it sounds if you come back to it three or four months later.

[kelldammit]

lol, i like the eno quote. yeah, i didn't go into mastering, as i was going more along the lines of the final mixes to deliver to mastering...(even if that means just dragging it over to another program). the idea is to start the (pre)mastering process with as good a mixdown file as possible.

[chrispick]

lol, i like the eno quote. yeah, i didn't go into mastering, as i was going more along the lines of the final mixes to deliver to mastering...(even if that means just dragging it over to another program). the idea is to start the (pre)mastering process with as good a mixdown file as possible.

Yeah, I wouldn't concern myself with mastering tasks except that my job kind of demands it.

The deal: I can't afford to send files off to an mastering engineer because the "final" version rarely ends up being the actual final. That is, I've no way to budget jobs accordingly. Nine times out of ten, I'll deliver audio for a hard, fast, drop-dead deadline only to get a call two days later saying the client shifted some dates and now wants another go-around with some minor tweak (and they need it this afternoon).

Moreover, I often find out half-way through a job that I'm competing against some cut by, say, Foo Fighters or Crystal Method (placed in as a temp track by some editor), so volume and fullness factors in quickly.

So, I've had to learn how to make my mixes sound as mastered as I can. It's been an uphill climb for sure.

If I were to make an album though, I'd definitely ship my final mixes to a mastering house.

[mrgkeys]

Does anyone know if DP or anything else supports Tiger's 64 bit Core Audio Format file?

[David Polich]

I don't think any Mac-based DAW's support 64-bit processing at the moment. I know Sonar for the PC does...

Hey Chrispick, I hear ya. When you're doing music for commercial purposes, tv, radio, movie soundtracks, of course you have to get a handle on mastering - not that we dare to call ourselves mastering engineers, but deadlines and economics kinda force the issue.

I actually re-mastered my band's album after an unsatisfactory visit to a mastering house. I just didn't get the results I wanted from the mastering house and couldn't afford to go back again. At least at my home studio, if something is just too abrasive or un balanced after applying mastering, I can go back to the DP project immediately and tweak it and then export the fixed file back out for another shot at mastering.

[chrispick]

Hey Chrispick, I hear ya. When you're doing music for commercial purposes, tv, radio, movie soundtracks, of course you have to get a handle on mastering - not that we dare to call ourselves mastering engineers, but deadlines and economics kinda force the issue.

Exactly. Goes with the job.

I actually re-mastered my band's album after an unsatisfactory visit to a mastering house. I just didn't get the results I wanted from the mastering house and couldn't afford to go back again. At least at my home studio, if something is just too abrasive or un balanced after applying mastering, I can go back to the DP project immediately and tweak it and then export the fixed file back out for another shot at mastering.

That, I think, is one of the cool things about it. I find that the more I learn about home mastering, the more I learn about mixing. And I have a lot to learn in both categories.

[archtoper]

Does anyone know if DP or anything else supports Tiger's 64 bit Core Audio Format file?

Ozone 3 puportedly processes everything at 64 bit. It's a pretty amazing little plug- especially when applied to a 32 bit source file