TPDF vs. POW-r Dithering

[Loudnoize Ent.]

Anyone use TPDF dithering and how does it compare to POW-r?

Thanks...

[Monte McGuire]

I personally think POW-r sounds better than simple TPDF for "large" dithering, such as going from 24 to 16 bit where you have a chance of actually hearing it. For situations where I'm dithering to 24 or 20 bit, usually in the middle of some sort of operations between processors, I'll use unshaped dither (essentially TPDF) just to reduce the amplitude of the dither but to make sure it's sufficient to avoid nonlinearity. For me, this seems to work the best.

For example, I usually follow L2 with a POW-r insert set for 16 bit. I set L2 to unshaped 24 bit dither and let POW-r go to 16 bit. Back when SpectraFoo was working on my rig, it seemed that L2 behaved better at microscopic signal levels with unshaped 24 bit dither, so I decided to stick with that. (Now that there's an OSX version of 'foo, I'll have to revisit this!)

As always, take a listen and see if it matters for you or your material. Sometimes, different dithers can impart a very subtly different timbre to a mix, so it's worth choosing a dither that matches the timbre you're aiming for. Might as well line your ducks up in a straight line... maybe it mattters, but it probably won't hurt.

Specifically, POW-r type 3 is more apt to lend a glossy, focused or 'hard' presentation, which could go well where you're trying to make something glossy and bright. Type 1 is a lot more gentle, so it seems to work better for me when i want the result to be less hyped.

I do find POW-r to be preferable to any other dither for going to 16 bit though. It really is good...

Best of luck,

-monte-

[Loudnoize Ent.]

Thanks for your response Monte. Out of curiosity, would you sometimes use Digi's POWr dither or 3rd party software, that is if not sent to a mastering house?

Thanks again.

[Alécio Costa]

I totally agree with MCguire. Power level1 more subtle, Power Level seems to reveal a hyped top end.

BTW... do you guys bounce to disk, bounce internally with bussing or mix to an extrenal 2 track machine (analog/digital).

I am a little off topic but more and more guys are repoting some "frying high ends" after ordinary bouncing.

Thanks

[minister]

i am have been reprinting more and more and bouncing less and less. even on simple things. i have begun to dither and SRC in BARBABATCH.

i have been messing with the Dither on the Sony Limiter. it has Type 1, 2, 3, 4 & TPDF. they have their different applications. TPDF, for me, does not work on just voice. Type 1 does. but i got TPDF to be the preferred method on a music mix. it is nice that you can adjust the percentage of shaping from 0-100%. a lot about the dithers are explained in the sony manual.

as for POW-r, i roughly agree with what has been said. i rarely use 2 because i often find that the high end sounds a little harsh. 1 is OK, 3 sounds best to me on most things.

...but it depends on the material...

[Jeezer]

Old post but in 2019 there's no difference between bouncing to disk or printing to a track in PT12+ at least...

As for the Dither question, for me.. POW-R2 for most things, but it can be program dependent.

TDPF = slight smearing and coldness (reduced bass, hyped top)
POWr1 = increased bottom reduced top - more mushy
POWr2 = the closest to source and any minute change is 'for the better' but really... hardly there if at all
POWr3 = Reduced bottom increased top (opposite of POWr1) mix becomes lopsided and hypes the high freqs.

Given all this, POWr2 seems the best if you want it as close to the Pro Tools master as possible. You may want to use 1/3 if you find it somehow improves a less than already perfect master.

[JFreak]

Just for kicks, try the Apogee UV22HR dithering. Easiest to buy Logic and dither there. I have no idea why there is no AAX plugin but is included in LX anyway. Much better than any POW-r version out there

[Bob Olhsson]

The real-world test is what sounds best after being converted with a lossy encoding codec or adjusted with today's digital volume controls. Almost nobody outside a studio is ever going to hear our dithering choices and theoretically, TPDF accepts additional processing better than anything other than floating-point files.

[JFreak]

Sad but true