Benefit of recording 32-bit audio?

[mbr]

http://www.youtube.com/watch?v=kF9cGCO5 ... ED225D584F

[Jarno]

http://www.youtube.com/watch?v=kF9cGCO5 ... ED225D584F

OMG! What a sh*tload of misinformation. The author doesn't understand anything about digital audio.

[iBM]

http://www.youtube.com/watch?v=kF9cGCO5 ... ED225D584F

OMG! What a sh*tload of misinformation. The author doesn't understand anything about digital audio.

I have to agree. He has no clue about the topic.
He has gathered all the "internet myths" (read misinformation) of digital audio in two short videos.

The worst part is that he probably got paid to do this videos (Lynda.com). I also noticed it was from 2007, and it just makes it OLD misinformation.

This is how, what I call Internet myths live on. Please someone remove this from youtube.

[Early21]

Have to thank all the experts who provided information here... I was not aware that I should turn off Hyperthreading... not the subject of the original post, but highly useful ( and mentioned in Steinberg Knowledge Base). Helped a lot with my performance issues with Studio Drummer, my go-to drum set.

Also, love those videos from Xilph - Monty Montgomery - very well explained.

Thanks, all.

[peterandu]

There are 2, less processing as Cubase works in 32bitfloat internally. Huge headroom, you don't have to worry about additive EQ clipping the signal, when going to disk.
I do mostly rock-pop stuff, where I don't use excessive processing and 24bit is more than fine. The file size in 32bit float is not that much bigger, as long as I can run 100-150 tracks I don't mind, any 7200rpm disk should do that. To be clear there should not be any audible difference between them.

Is there technical manual about the headroom and bit depth? from cubase, i mean.

[peterandu]

Here's a clear example:
Take an audio recording.
Push it really high using event volume.
Bounce the selection.
Now drop the event volume back down.
If you were in 24-bit, you'd see clear clipping and distortion.
With 32-bit, it returns to the correct levels, not losing whatever went about the threshold.

If i creat a empty project with 32 floating bit depth setup and import 24 bit audios for mixing. What will happen, will the 24 bit audios be converted to 32 bit format?

IF not, does it mean the extra depth is used as a virtual headroom?

[peterandu]

And yes in that context you can hear a diff.

{'-'}

You are imagining things. In a properly executed blind test you will NOT be able to hear any difference. You are "hearing" a difference because it is a bigger number of bits and you want to hear it due to the psychological bias you have. Also, if you are not doing any offline processing there is ZERO difference between recording 24 bit fixed and 32 bit float.

The only way you will see a difference is to intentionally force errors using offline processing as Soul-Burn described, but if you have anywhere near remotely decent gain-staging habits, 24 bit fixed is more than adequate resolution and the mixer is always floating-point no matter what file type you use.

Last, with modern computers I do not believe you will see any "less processing" load on the computer by recording 32 bit float files. The files are bigger too, and if you record much this can be a penalty (despite what others may claim). I record many tracks every day and back them up after every session. File size is still an issue for me even with today's large inexpensive drives.

What happened a 24 bit depth audio is imported to a 32 float project for mixing? Will the audio be converted to 32 float F
format? Is the extra bit depth used to add extra headroom to emulate the analog gears.

[MattiasNYC]

Is there technical manual about the headroom and bit depth? from cubase, i mean.

Not sure you'll find it. There's enough headroom within Cubase and Nuendo that you don't have to worry about clipping before the outputs. You still have to make sure you don't clip the outputs (going to fixed point converters and exported files stored as fixed point).

It's still valuable to keep levels below 0dBFS throughout the signal flow even if you are in floating point processing DAWs.

If i creat a empty project with 32 floating bit depth setup and import 24 bit audios for mixing. What will happen, will the 24 bit audios be converted to 32 bit format?

I think you are asked about that when you import your files, or there is a preference you can set for it. Either way you don't really have to worry about it unless your computer is pushing the very limit of how it performs. Because going from 24-bit audio files to the internal 32-bit floating processing (or now 64-bit float) doesn't seem to be that big of a deal unless you're really pushing your computer.

So, if your recorded audio is 24-bit fixed I would just keep it like that, personally.

IF not, does it mean the extra depth is used as a virtual headroom?

No, you're thinking about it the "wrong" way I think. There's an article on Wikipedia on Floating-point arithmetic that may help, but it's a bit technical of course. There isn't really a particularly good and easy way of explaining it.

Basically, Cubase's internal signal path is always going to be 32- or 64-bit floating point processing, so any calculations done have this enormous headroom regardless of whether or not the audio originated as 8-, 16-, 24- or 32-bit fixed (or float) point audio. The "extra depth" isn't really about the number itself in this case it's about the ability to process large numbers.

Is the extra bit depth used to add extra headroom to emulate the analog gears.

Cubase's and Nuendo's basic signal paths dont emulate analog at all by themselves. You need to use plugins to emulate analog. If you use a plugin to emulate analog that still has nothing to do with floating point versus fixed point processing.

[mroekalea]

The 32 bit float format is a mixer format, that is the mixer runs in 32 bit float, your audio files will remian 24 bit.

[Split]

unless you use a AXR4T

[GlennO]

The 32 bit float format is a mixer format, that is the mixer runs in 32 bit float, your audio files will remian 24 bit.

You're confusing the processing format with the recording format. You're talking about the processing format, but this old thread is about the recording format. There is a 32 bit float option for both, but they are unrelated.

[mroekalea]

The 32 bit float format is a mixer format, that is the mixer runs in 32 bit float, your audio files will remian 24 bit.

You're confusing the processing format with the recording format. You're talking about the processing format, but this old thread is about the recording format. There is a 32 bit float option for both, but they are unrelated.

Found this: New 9.5.40.update says: It is now possible to record audio files with a 32-bit integer bit depth and also export audio in 64-bit float and 32-bit integer.

Is seems that this is since version 9.5.40 a possibility. So my statement is/was true, only for versions below 9.5.40

link: viewtopic.php?t=145759

[MattiasNYC]

The thread is like half a decade old, with the exception of the guy who asked a follow up question which was answered...

Just in case someone missed it.

[GlennO]

The 32 bit float format is a mixer format, that is the mixer runs in 32 bit float, your audio files will remian 24 bit.

You're confusing the processing format with the recording format. You're talking about the processing format, but this old thread is about the recording format. There is a 32 bit float option for both, but they are unrelated.

Found this: New 9.5.40.update says: It is now possible to record audio files with a 32-bit integer bit depth and also export audio in 64-bit float and 32-bit integer.

Is seems that this is since version 9.5.40 a possibility. So my statement is/was true, only for versions below 9.5.40

link: viewtopic.php?t=145759

Now you're confusing 32 bit float files with 32 bit integer files . Cubase has supported 32 bit float files (which is what this thread is about) for many years. Processing format and file format are two different and unrelated things in Cubase. Neither one affects the other.

Anyway, I'm pretty sure the person who resurrected this ancient thread had their question answered already.

[Nickeldome]

Aloha guys,
Yes yes yes yes!

Here I do mucho local stuff that many times is one
person singing (chanting) or one
solo flute type instrument or
a single slack key guit or uke.

And yes in that context you can hear a diff.
Especially with a single vox with no FX.
And at my age too!

{'-'}

This is called 'psycho acoustic perception'. You expect 32 vs 24 bit to sound better and because you determine the numbers look better you will fool yourself and conclude it actually does?

I remember a massive hifi test back when FLAC was just introduced in the mid 2000's on the market. They let the subjects ((known hi-end owners and experienced critical listeners) listen to varying sources like classical, jazz and acoustic music from MP3 96~320, FLAC and WAV (cd) on a top notch high end system.

The result was that everyone was clearly able to point out that everything below 256 kbps was clearly inferior quality. But after that it got undecided? Some people even preferred MP3 320 kbps over FLAC or WAV?

So if you can really hear that 32 bit floating at 192 khz (or even at 16 khz for that matter...) sounds better than 24 bit you must truly be one of a kind!

When we're talking about mixing and headroom it 'may(?) 'be a different story? It never hurts to start at the best possible resolution because you may loose quality in the (analogue) process? But then again? Looking at the test above? Would it really matter in the end?

And lets be honest? We start mixing at the highest possible quality and then we decide to apply 'tube, tape, vinyl plugins to give it that 'analogue' feel to degrade everything we started out with?

So maybe we're just fooling our self's in pursuing this maximum quality?

In the end we don't want to hear 192khz/32 bit? But rather a sound that maybe resembles a plain analogue 20hz~20khz signal?

So the real question is? Why mix at incredible high resolutions when we know that almost every one wants to hear that familiar 'lower-fi' sound?