What's Your Preferred Output Format?

[FMiguelez]

Yup. Here it is:

http://www.motunation.com/forum/viewtop ... 60#p357707

Notice the date of the thread... And we still don't know this FOR SURE by now????

- Is it even working at all in DP9?

- When is the manual going to be corrected so it explains accurately what really happens? The glowing error and misinformation has been in every manual since at least DP4!

- Is recording a MAS or a DP function?? (as in printing a submix in realtime)?? What happens with the conversion in this case?

Ok. I better open a new thread with that explicit question, because now I REALLY want to know the answer to that.
If you're interested>
http://www.motunation.com/forum/viewtop ... 26&t=60520

[EMRR]

Interesting old thread there, thanks. Raises a lot of questions.

24 bit ADC into DP at any bit rate. Is the ADC applying dither? Does it depend on the ADC in question? Is DP doing it if you record 16 bit from a 24 bit ADC?

What differences exist with a 24 bit mix bounce as opposed to a single channel 24 bit output to a DAC which has had a gain change or processing applied? The mix bounce is a first generation 24 bit capture like the ADC input question above, while the single channel manipulation is....a manipulation.

I didn't quite say it, but are the 32 bit input DAC's in the AVB line applying dither when their gain trims are changed, since they are NOT 32 bit output?

I see Audio/MIDI describing the 16A as 32bit float with DP8 set for 32bit float recording.

Spectrafoo will tell a lot about it, and tighten up the set of questions. At least turning on dither in the plugs is easily visible in the noise spectra, along with noise shaping.

[Shooshie]

b-wave 48/24 interleaved

Ditto, with brick wall limiter, dithered with noise shaping. The limiter may raise the volume slightly, but not by much; it's really there for transient security. I try to leave 6 to 8 dB headroom over the highest transients, but sometimes one gets through. Not with a brick wall.

Shooshie

[Rick Cornish]

Ditto, with brick wall limiter, dithered with noise shaping. The limiter may raise the volume slightly, but not by much; it's really there for transient security. I try to leave 6 to 8 dB headroom over the highest transients, but sometimes one gets through. Not with a brick wall.

Shooshie—what are you using for a brick wall?

[Shooshie]

Shooshie—what are you using for a brick wall?

Usually Waves L3 Multi-Maximizer.
I've been waiting on Slate's FG/X rewrite to be finished, at which point I'll probably resume using it again. It was an amazingly good limiter when I was using it before, but that's been a while and now I don't even remember if it was bricked or not.

The way I see it, when you're not using the limiter for a lot of sound shaping and beefing up the volume, but you're mainly using it for stopping any unruly peaks, it's not so important which one you use if it does the job. But when I AM shaping the sound, the L3 MM is truly a great limiter, but the FG/X is magical. And both are tricky to use, but that's part of doing this job well.

Shooshie

[monkey man]

I'd like to see a good argument for switching to 32bit floating-point from 24 bits... at any stage of a project.

Not saying that there isn't one, just that if there is, I'd love to see it.

Dont know who or when this was posted, as I had collated it into a note sometime ago. But here...

"• 32 bit floating point audio file format - This is only used by some high-end applications, but if you are bouncing audio from DP, such as consolidating tracks into stems, or freezing a MIDI VI, you can use this format to keep your audio consistent with DP's internal processing. In other words, your audio tracks can be printed at 32 bit floating point so that when you go from internal audio processing in DP to an external file that may be imported back into the app, you don't lose any detail whatsoever. With 24 bit audio, DP has to truncate the samples when bouncing to disk. Re-importing then means adding zeroes, so the digits they replace in the original (pre-bounce) are simply lost. 32 bit FP, however, means your audio will never clip until transferred into a less detailed format.
=========================
The benefit of using "32-bit floating point" files comes from not needing to convert the samples to/from the internal representation used by the host when communicating with the storage medium. No gain changes (normalization) are needed, and no rounding errors are burned into the file. This comes at the cost of a 33-1/3% increase in disk usage, a paltry sum by modern standards. It doesn't make any different for raw takes recorded using currently available converters, but it can make a difference for stems, printed VIs and printed effects."

Labman, thank you for responding to my question.

Call me non-audiophile, but I frankly don't care about rounding errors if they're 1500dB down, which, if I'm not mistaken (and I probably am!), is the theoretical noise floor for DP's internal processing of 24bit audio.

I therefore remain unconvinced, and my original "challenge" still stands.

EDIT: I'm not trying to be rude; it just came out that way, mate. Sorry!