Exporting 96 khz/32 bit float from Audacity

[WarmJetGuitar]

How do I export 96 khz, 32 bit float files from Audacity to Pro Tools?

The thing is that for some reason the WAV-files created by Audacity is only 16 bit which is kinda substandard as we track to tape and release our music on vinyl.

Any advice would be really amazing. How do I cope with this without degrading the sound?

 

[Bobbsy]

I'm not an Audacity user but can you not just "Save As" the necessary files onto an external HDD, thumb drive or whatever rather than worrying about Export commands?

 

[Ethan Winer]

for some reason the WAV-files created by Audacity is only 16 bit which is kinda substandard as we track to tape and release our music on vinyl.

If you compare fidelity specs, Wave files at 44.1 / 16 are much cleaner than analog tape and vinyl. The frequency response is also much flatter. So exporting to standard Wave files will not degrade the sound.

--Ethan

 

[bouldersoundguy]

Given that mega hit recordings have been done entirely in 44.1/16 I can safely say that 96/32-float is drastic overkill for what you're doing. I like 48/24 for most things and even that's way beyond adequate. If your project is in 96k then maybe keeping the mix files at that sampling frequency makes a little sense, but at least skip the 32 bit float and export mixes at 24 bit. The 32 bit float setting is more of a DAW internal processing thing.

 

[Bobbsy]

Are we talking about exporting mixes here or individual stems/tracks for mixing in Protools?

If they're mixed tracks then I fully agree with what the others have said. If it's individual tracks though, I must admit I always mix in 32 bit float (though 44.1 or 48 sampling, depending on whether it's a CD or video project) and convert the bit depth after the mix.

 

[miroslav]

The 32 bit float setting is more of a DAW internal processing thing.

Right.

My DAW is always working at 32 bit float internally...but when I'm exporting files, even to simply move them to another system (occasionally), I use 24 bit. For final public consumption, it's 16 bit, of course.

AFA what's "best" for frequency and bit rate....choose what makes you feel warm-n-fuzzy.
There's really no reason NOT to use higher numbers if your system can handle it.
Some people will say, it just might "future-proof" those files (if that's important to you), as there are still some unexplained oddities with audio/sound/hearing that need more research....though I will agree that for most cases, especially on the home-rec front, whatever is turned up by science at some point, is probably going to be too fine to matter for "Rock-n-Roll"....
...but again, that's no reason NOT to pursue it if it makes you feel warm-n-fuzzy.

 

[WarmJetGuitar]

I can hear the difference between 44100/16 and 96000/32 bit float. Probably 32 bit float is overkilll compared to 24 but as we'd started doing mixdowns that way we better stick to it to avoid any complications. But overkill to use higher than 16? You must be joking man.
Don't tell me that a Studer A800 don't have a wider frequency range than a CD - not that we're working on a Studer, just a cheap but nice sounding Fostex. And yes, there might be top selling albums recorded at 44100/16 - but there's also lots and lots of top selling albums that sounds like shite in regards of sonics, production and recording tecnique. Cleaner ain't always better.

Even if I couldn't hear the difference between different bit and sample rates I'd still go for the higher spec as any AD/DA conversions or use of plug-ins during mastering degrades the fidelity - and hard drive space is really cheap.

But well... to the original subject: the "save as..." function doesn't do the trick. If we end up getting the mastering done digitally I'd love to convert the file to a format useable in Pro Tools or similar without degrading the sound. Anyone knows how to solve this?

 

[bouldersoundguy]

I can hear the difference between 44100/16 and 96000/32 bit float.

There are several ways there could be a difference between them, not all of them equal. Exporting 44.1k audio from a 96k project involves resampling. That would be different from comparing them all the way through the project. Some converters simply sound better at one or another sampling frequency, and not necessarily the highest one.

Probably 32 bit float is overkilll compared to 24 but as we'd started doing mixdowns that way we better stick to it to avoid any complications.

What complications? 24 bit is already overkill and 32 float is way overkill. 16 bit is quite adequate if you're careful with your gain structure while 24 offers some extra of S/N so you can be a little lazy about levels. There is absolutely no advantage to 32 bit float mix files since it's all going to get truncated anyway. The advantage to 32 bit float is inside the DAW where all sorts of processing is going on, and most DAWs do that automatically.

 

[Bobbsy]

32 bit float is NOT overkill during the mix process--but that's not because it sounds better. It's simply because the insanely wide dynamic range gives you lots of room to take the easy way out and not worry about levels while you play with the mix. Even if things get way too hot, pulling them back down to appropriate levels gives you the same sound clip-free. Of course you can achieve the same things with integer files but, just as the advice is often to "listen to the mix, don't worry about numbers" 32 bit float is just another way to achieve this.

As you say, many/most DAWs work this way anyway. I'd normally prefer not to have conversions of bit depth going on behind my back but, frankly, in Audition at least I can't hear a jot of difference whether I record that way or just let the DAW play around internally.

Anyway, back to a previous question: at what stage in the process is the OP trying to do this. If it's final mixes, I can't see a single reason to try and keep things at the higher bit depth/sample rate. However, if he's transferring tracks and stems to finish the mix process in Protools, maybe--I'd rather keep all my conversions to once at the end of the mix process.

 

[bouldersoundguy]

32 bit float is NOT overkill during the mix process--but that's not because it sounds better.

Right. I was saying it's overkill for the exported/bounced/rendered mix files. During the mix process 32 bit is pretty much industry standard while 24 bit is the standard for storage and transfer.

 

[Farview]

Another problem I've noticed with 32 bit float files is that there seem to be compatibility issues between programs. It's as if they don't all use a standard format.

32 bit float os just 24 bit audio with 8 empty bits at the end that get used during processing, only to be discarded on the way to the da converter. Great idea for processing, absolutely no point to it for file transfer.

 

[Bobbsy]

I've often wondered if different DAWs implement floating point working in different manners.

In Audition (the only one I have an in depth knowledge of) 32-bit float is a special form of 24-bit recording which is lossless even when severely processed - it uses scaling bits to store the amplitude of the signal, and has a much larger range than 24-bit integer recording. But essentially it is 24-bit - just stored in a slightly different way. Basically, it has a 23-bit mantissa + sign bit + 8-bit exponent, which is effectively 24-bit. The reason for working like this is to retain the resolution when audio has any sort of amplitude-related operation carried out on it - which it can do over about a 1500dB range. If Audition worked in 24-bit integer mode, this range would be restricted to about 144dB.

So, I wouldn't exactly call them "empty bits" because they're darn useful--but it's also why I can swap between 24 bit integer and 32 bit FP with no fears about losses in the conversion. However, I really don't know if other DAWS claiming to have 32 bit FP working handle thing the same way.

 

[Farview]

That is what they all do, but when the actually write a 32 bit float file, it seems there are differences between manufacturers. Last time I tried, a 32 bit file created in soundforge would not open in wavelab. ( might be the other way around, it was a long time ago). And just about every time someone sends me a song to mix with 32 bit files, there is a problem opening them.

 

[Bobbsy]

Certainly the Audition user forums tend to recommend using 24 bit integer (or whatever bit depth you started in) as the exchange format and leave 32 bit FP as the internal "working" format.

 

[Ethan Winer]

16 bit is quite adequate if you're careful with your gain structure while 24 offers some extra of S/N so you can be a little lazy about levels.

Exactly. As they say, you can lead a horse to water but you can't make it drink. I think the audio corollary to that is you can explain the facts, but you can't make them understand or accept them.

--Ethan