Guys, I always work at high resolution as possible. I need some conversion support for posterior work, but have no time to googling.
Need a pro editor or converter just as a converter step only. First, it must open/import/support 64 bit float, and then export support for 32 bit integer / fixed.

Rendered source, import support: 64 bit float.
End format (save as) target: 32 bit fixed(integer).
Sample rates: 44.1 - 192 kHz (and / or higher, but not mandatory).
Format: wav pcm stereo.
Extra, not mandatory: AIFF stereo etc.

No 32 bit float please.

P.s. RX doesn't support 64 bit float for import, but supports both 32 bit versions.

 

64 bit float? Woah,never even heard of that

 

Can't you do this natively in Reaper? Any program that can output at the specs you want and is able to load the file, I assume, would work. Adobe Audition, Wavosaur, Audacity. Unless it's some proprietary format like Wave 64.

 

that's why I ask. I use reaper. You can render either 64 bit float or 32 bit float. But izotope rx supports only 32 bit float or 32 bit fixed. I need 32 bit fixed to open in izotope rx after reaper render.

 

It's all about the headroom. Just adjust your levels, so that you don't need 1000 dB anymore, and render to 32 bit float that you can use in RX.

 

Yes, thanks, I work this way, but need 32 bit integer fixed for less noise floor. I usually at mastering do render without limiters, then posterior rx editing, then limiting in Reaper. So needed less noise floor while editing non limitered audio. And this is also for final high resolution 24 bit wav.

 

24bit has a noise floor of ~170dB. That's enough.

 

So, no need panic from me) thanks, maybe it is useless, but for some other material it would be not so bad to have this type of workflow.
So, 32 bit float is enough. Solved.

 

just for the papers
32 bit ~192.66 dB SNR
24 bit ~144.49 dB SNR

that is enough like allready said.

 

192 dB SNR for 32 bit integer fixed and 64 bit float. 32 bit float has a precision of 24 bit, as I measured and FabienTDR from Tokyo Dawn Labs said on gearslutz. As I measured test sounds from src.infinitewave.ca, 32 bit float has the same noise floor as 24 bit does, but a couple dB lower. 32 bit integer fixed has much much less noise. So, 32 bit fixed is better. Float is only to preserve 24 bit audio from clipping above 0 dB. 64 bit float is to preserve 32 bit int/fixed from clipping etc. So if my target final format is 24 bit, I need 32 bit fixed for RX works. Because RX doesn't support 64 bit float.

 

24 bit noise is more than enough to make you loose your hearing many times before you can hear quantization errors or even the dither which is greater than the noise floor by design.

Unless you are doing scientific experiments at very high SPL you will be more than fine.

 

https://www.quora.com/What-is-noise-floor-in-terms-of-audio
http://www.ittc.ku.edu/~callen/energy_harvesting/Meloy2003pp12.pdf
http://www.vlf.it/naturalnoisefloor/naturalnoisefloor.htm

 

So I don't need worry about higher resolution) it is just my confusion to make things maximally better, without seeing that's is absurdly too high, because less is already enough.