I’m curious to know, who is taking advantage of 96000, and why you are doing it, if you are… I’m also wondering what dither type, Renoise does, when you select dither in the preferences… Right now, I mix @ 44100, with 32Bit IEEE Float, and dither is turned off. When I render my mixes, its @ 32 bit… and then when I master, and convert to 16 Bit wav, I turn the dither back on.
What do you guys do?
Cheers
I basically work in 24 bits 96Khz with undithered output. And I often start to export my track with the realtime mode so that I simply have absolutely no surprises about it. What I stopped to do is : exporting the music with a lower sample rate, because some DSPs do not behave exactly the same when you use for example 44KHz instead of 96Khz, especially the IR-like based devices like the *cabinet simulator.
However, I sometimes take the time to test all the other available output render modes, and alternative output parameters, when it’s export time. We never know. I.e : with dithering, without dithering. With soft clipping, without soft clipping, with Arguru’s sinc, without it… Sometimes, even if the realtime mode sounds exactly like I want, the Arguru’s Sinc interpolation looks sweeter on some high freqs, and that’s even better, so I keep it, it’s sometimes “the final good surprise” of the process. But testing all these alternative renderings could be long. So, if someone could make a tool, you see, that launch in one click several versions of the same song, with all the available rendering modes, so that I start it, prepare a cup of coffee, think to something else, go outside, and come back home, and get all the possible rendered versions of my songs : it would be cool. 
Interesting… So if I am working along @ 44100, and I decide at the last minute, “after I have completed the composing and mixing,” to up my rate to 96000; I might be in for a big surprise? My mix might not translate? So if one was to do 96000, they had better start from the very beginning and do the entire project at that rate…
Hmmm…
I also find it interesting you are using 24 bits… I actually, “hear a difference,” in 32 bit. Maybe that is psychological, I don’t know… But I sense a, “more full bodied sound, and great depth,” for lack of a better explanation…
I’ve never used Argurus sync… I just use, “Priority High and Cubic.”
Lots to think about, and you are totally right, to render every possibility would be quite time consuming…
Thanks KURTZ, cheers
I think you’re confusing the floating-point processing (32 bit) with the actual rendering bit-rate (24 bit). AFAIK 24-bit is still the highest bit-depth widely available for digital audio as 32-bit DACs have yet to be developed for general commercial use.
I guess that’s the general rule of thumb: your source material dictates the quality of your final output. I generally render at 24-bit / 48kHz and don’t really hear a lot of difference between this and higher sampling rates, but I can definitely hear a difference between 16 and 24 bit rates.
Woah!! Hold on, really? These settings here, render a 24 bit wav file? You are right, I am like totally confused
Aww sheeeeeiit, that must be a new to 2.8 or sumfin. Apologies, my bad; I’m still using 2.7 on a knackered old G4 powerbook and haven’t seen that option before - count me in for a big old slice of humble pie!
I guess it’s great it if you’ve got 32-bit converters, but if you haven’t it’s gonna come out the monitors at 24-bit anyway.
I wouldn’t worry about apologies, 
I make a zillion mistakes… but what’s really odd, is that now I have noticed… none of my other daws are using/producing 32Bit wav files… So now I am just totally blown away…
In other news, I just tried Argurus Sync for the first time, and imo, it sounds fantastic… I also rendered at 96000 for the first time… and I have caught, “a surprise.” I feel like my track, is a bit more bass heavy, than what I was hearing… which was, “not at all bass heavy.”
The other thing I noticed… which is not a problem, but kind of a problem lol!! If you want to take advantage of this new, “32 bit Rendering Feature,” you must master in Renoise, or in a DAW that supports the 32 bit wav file… and like I said, I just checked all my other daws, this 32Bit thing seems new and, “Renoise only.” ( It could also be in Cubase or Nuendo, I don’t know, as I don’t have those ) So… what happened when I loaded the 32bit wav into another daw? It got resampled… I do not know if this adds a dither type, but if you convert the resampled wav, to 16 bit, and add a dither type, you might actually pass the file twice with dither!! Which is no good!
I have been mastering in Renoise anyways, but now I have a mastering project ahead that concerns, “more than 1 track at the same time.” Which presents a set of issues… but I’m thinking up a game plan… because I want to take advantage of 32 bit… I hear extra depth! I don’t want to apply dither to 24 bit, and apply dither to 16 bit, and I certainly don’t want, “automatic,” resampling… Who knows what that does?
Weird… hmm… Very interesting feature though
Check out the wav form, it looks like, “candied sugars,” and this is not going over zero, btw… its just, “zoomed in.”
About the Renoise 32 bit format.
It is really 32 bit float WAV file. Which is slightly different from the other two formats which use integer samples. The benefit of using floating point is no loss of quality across the dynamic range. Very quiet material doesn’t lose quality as the precision remains the same regardless of volume.
Anyway… The format is nothing new. And all audio programs in my computer can handle it just as well.
(P.S. I am not 100% sure if it is limitation of Renoise or WAV format, but 32 bit format should also allow louder than 0db sounds, which on normalization in other DAW should become non clipped. This doesn’t seem to be the case right now, renoise exported 32bit files are hard limited to 0db)