Dither When Mixing Down to 24 Bit

[Adler]

Hello all,

When working with a 24 bit session and mixing down to a 24 bit file, do you find yourself applying dither to the mix buss? I am just wondering if this is a process I should follow since Pyramix does its internal processing in 32 bit floating point.

Any input would be appreciated.

Chris

[Paulo M]

Hi,

I only apply dither when mixingdown to 20 bit (digibeta) or 16 bit files. If you rrecordings are 24 bit and the end files also 24 bit there is no need to dither.

[Perfect Record]

I only apply dither when mixingdown to 20 bit (digibeta) or 16 bit files. If you rrecordings are 24 bit and the end files also 24 bit there is no need to dither.

Hmmm- don't know if I agree there. If I've applied any processing, even a trivial level change, I always dither, whether 24 b or 16b. I often dump 32 bit files of mixes, then dither externally using the dither flavor of the day, then bring that back in for the final CD assembly.

[Adler]

Thanks for the responses so far.

I never really thought about this issue until I noticed a subtle difference in sound when comparing the output of a multitrack project (24 bit, 44.1 kHz PMF files) to the mixdown of said project (24 bit, 44.1kHz undithered WAV file). I'm not saying there was a glaring difference, but the stereo image of the mixdown seemed more closed than the multitrack. Could this difference be attributed to the lack of dither on the mix buss? Or I am just fooling myself into hearing something that isn't there?

The irony of the situation is that I never got in to the whole auditioning different types of dither thing - I usually just default to TPDF with no noise shaping when creating CD masters (although I'll admit that HDCD sounds really good, and I'd like to try POW-R). And to think now that the application of dither to the mix buss may (and I stress, may) improve the quality of my mixdowns. Or this could all just be in my head.

Chris

[fl]

I suppose it all depends on what you're likely to do with your mix after you've made it. If your intent is to do additional processing - perhaps to drop down to 16 bit for CD - then the dither I'd use for any initial mix to 24 bit would have to be flat frequency response - no noise shaping. I'd only use a noise shaped dither - be it POW-R or any of the stock shapes offered by Pyramix - when I know that the result is going to be the absolute, final form for the material.

Lately, my noise-shaped dither of choice has been the M-BIT+ dither by the Izotope folks, so in Pyramix, I'll mix down to 24 bit in whatever sample rate the session is in, and then do the additional processing (SR conversion, word length reduction) in Wave Editor to make use of the Izotope SRC and Dither algorithms. As a result of this thread, I may just start dithering to 24 bit with flat dither in Pyramix, as opposed to my current practice of applying no dither when I'm mixing 24 bit source material.

Since Izotope offers to license their processes to other parties, I'd sure be in favour of Merging incorporating their Dither and SR Conversion into Pryamix, as it would certainly streamline the whole procedure.

[Thomas Grubb]

Hi Chris,

Is there any particular reason for mixing to a 24-bit file (apart from bouncing to a 24-bit external mixdown machine)? I do all bounces to 32-bit floating files just to avoid the whole dithering issue. This also keeps things simple for final CD mastering where you inevitably need to change the levels of a track or add fades before creating an image. It also keeps things a little more "future-proof" I think by mixing to 32-bit floating files (unless you want to open them in ProTools again I guess!).

I also like the flexibility of Ozone and the dithering options, though I must say I've never noticed much difference between different dithers (except for a harpsichord project once where the addition of dither (UV22 in Wavelab) added a slight fuzz to the attack of each note - that project ended up being truncated to 16-bits, but don't tell anyone )

Cheers,
Tom

[Adler]

Hi Tom,

I guess mixing down to 24 bit is a force of habit learned back in the day when I actually used a console and external recorder. You're absolutely right about using 32 bit floating point files...I will give that a go.

As an aside, I usually spend next to no time thinking about dither, but I have listened to a number of different types and (with the possible exception of HDCD) prefer flat TPDF 99 out of 100 times. It just sounds more like the source to me.

Thanks again to everyone who has contributed.

Chris

[Perfect Record]

I fully agree with Thomas' comment about future proofing and using 32bit float. It's a good idea to put a copy on the shelf at as high res as you can make for future release possibilities.

Regarding dither types, as has been discussed here a couple times already, there is a very well presented dither comparison online at:

http://www.24-96.net/dither/blind_shootout.htm

Check it out. You be the judge.

And as I've already stated here before, just like Frank I'd love to have mbit dither in Pyramix. Mark W recommends using the Ozone plugin as a method of getting Mbit without going external to Pyramix.