To Dither or Not To Dither, that is the question

[MIDI Life Crisis]

[cbergm7210]

Yea, that's me on this topic, brother.

[ccrane]

I have always left dithering off for the fear of double dithering, and always worked in 24 bit mode (DP5.13).

After reading this thread many times over, I feel it's almost inevitable that I've done something wrong at some point!

If I'm in a 24 bit session, and I bounce down to a 24 bit file DP apparently does NOT add dither as bouncing is a "MAS" function.
My question is have I just erroneously truncated a floating 32 bit session into a 24 bit file?
Magic Dave says "When rounding from floating point to fixed integer, dither is not required. "
So I take it that no dithering is needed on the master fader at that point.
It just never occurred to me to dither to 24 bits as I felt I was already working in 24 bit.

Another concern I have is that sometimes I will bounce down to a 24 bit file and reimport that file into a new 24 bit session and then perform some edits (cuts & bite leveling usually) as well as mastering tweaks to that imported 24 bit stereo file.
I often will do a file merge after I'm done with all of my edits.
It sounds like at that point, at the very least, I should have dithering on.

It's only after all of that I will finally dither down to 16 bits using a plugin on the master fader.

That dither setting has always caused me a bit of stress. Guess I have some good company.
It was all so much simpler when I just had my Roland MKII sequencer.

Peace,
Chris

[James Steele]

...whether 'tis nobler in the mind to suffer the slings and arrows of outrageous truncation...

[Umbrella]

Or to take arms against a sea of square waves, and by opposing, round them?

[newrigel]

Seems like he can hear the effect of dithering... and ONLY him!

[Kubi]

Yes, I mean the fact that this isn't easily audible (in which case we wouldn't have this discussion in the first place, it'd be obvious) puts the whole issue in its proper place, namely, there are best practices, but it's generally not a train wreck if you dither once too many or few...

Thanks for the post with SOS letter, the one pointing out the difference between convert and export: That's the key, I'm sure. Conversion obviously would apply dither, an export would only if it also truncates (i.e. exporting a 16bit file from a 24bit project.) This article adds another source recommending to leave dither ON: SOS, in addition to MOTU. Between that and the fact that gentle shaped hiss is always preferable to truncation garble right down the middle, that's good enough for me.

[Dan Worley]

Conversion obviously would apply dither, an export would only if it also truncates (i.e. exporting a 16bit file from a 24bit project.)

I'll just mention it again. As far as I can tell, testing DP 6.03 on my setup, dither is not applied on export at all (24-bit exported to 16-bit). I didn't test conversions.

c-ya,

Dan Worley

[Kubi]

yep, I should have written "should only if it truncates". I do my downsampling and bit-reducing in peak anyway...

[dopamine3]

After all that I'm confused about how I should handle the following:
I have a 24 bit project that was recorded in 5.13. I need to export the
audio files to .wav for the mix to be done on another system. I would
normally bounce all the SDII files to individual .wavs with dither off.
To complicate the situation I've just gotten DP 7 and I'm wondering if
it might be better to open the original 5.13 project in 7.2 and do the
.wav conversion in DP 7.2?? Should dither be off or on if I go that
route??

Thanks,
dope

[Kubi]

Turn it off for the procedure.

If you bounce there's no dither applied, so no diff if it's off or on.

If you export apparently there isn't either, but we don't know that from official sources, so turn off to be safe. In any case if you simply change an SDII to a WAV you don't want dither.

I'd always prefer loading the old file into DP7 and doing what needs doing there. I just always like doing all work in the newest possible version of software, since I'm not privy to all details that got improved from update to update. Plus in DP7 you could have DP do it automatically by setting WAV as the default format and turning on automatic file conversions. But for sound quality it doesn't matter what program you do it in.

[cbergm7210]

The one thing that is glaring to me at this point is the SOS article stating a double dither scenario that does not prove itself out when one tests that situation with inverting and canceling.

[Phil O]

The one thing that is clear to me is that this issue is still unresolved. Even the "experts" can't seem to agree. I'm thinking that if (and it's a big IF) you can hear dithering of a 24 bit file, it will be barely perceptible. And if you can hear truncation noise of a 24 bit file it will be just as barely perceptible. The difference at such a low level (what is it like -144dB or something?) between the two noises is probably insignificant.

What I'm saying is maybe you can hear that noise, but to be able to distinguish the difference between dither and truncation at -144dB is probably not realistic. I'm interested in the theory, but I ain't gonna worry about it, especially since I've been setting my default project format to 32 bit floating point lately.

Phil

[dopamine3]

Thanks for the helpful reply Kubi.
I was wondering if DP could do the .wav conversion automagically.
Sounds like the way to go!!

Thanks Again,
dope

[James Steele]

I've been leaving it off. So far nothing I've done has made me say "OMG! That sounds terrible!" So rather than hurt my brain trying to make sense of it I'll leave it set to where it's been set.