Dithering?

  • Thread starter Thread starter ClapHands
  • Start date Start date
Happy Holidays Glen and to all my friends at HR.

May the season be as warm as tube gear and the days be dithered with joy.
 
miroslav said:
Dithering is used (or not, it's a choice not an absolute) during bit rate conversion...like when going from 24 to 16.

You are already there...so no need to dither.
Click to expand...

Not correct. During mixdown, you are effectively going from greater than 16-bits worth of information down to 16 bits. Don't believe me? Mix together two sine waves at 0dBFS without cutting the levels and tell me you aren't clipping something fierce. :)

More to the point, no modern audio applications mix in 16-bit or 24-bit. They all mix in either 32-bit float or 64-bit double float. Therefore, every time you mix, you are doing an explicit bit depth conversion. I'm pretty sure you'll find that the same is true even in standalone hardware devices.

That said, since you're talking about a standalone device, I doubt you have any control over whether it does dithering, rounding, or truncation after the final mix stage.
 
dgatwood said:
Not correct. During mixdown, you are effectively going from greater than 16-bits worth of information down to 16 bits. Don't believe me? Mix together two sine waves at 0dBFS without cutting the levels and tell me you aren't clipping something fierce. :)
Click to expand...

Read the OP's question.
He's already at 16bits...I didn't see any mention of mixdowns in a DAW.
 
miroslav said:
Read the OP's question.
He's already at 16bits...I didn't see any mention of mixdowns in a DAW.
Click to expand...
That's how I read it. The Fostex is a DAW-in-a-box that records everything at 16-bit. It may do much of it's internal processing in 24- or 32-bit, perhaps, but if so, it's up/down converting at the digital I/O stage and outputting and saving at only 16-bit (to/from disc, as well as probably before any digital out ports it may have.) It may be internally mixing at higher than 16-bit perhaps, but the data going in is 16-bit at it's source.

And the example of the square waves is faulty because when you sum two signals at 0dBFS, it doesn't matter whether the data is at 16-bit or 256-bit; it's still the summing of two 0dBFS signals. You'll still wind up with an identical final sum of greater than 0dBFS. Whether it appears to clip at 0 or not depends upon the bit width of the summing engine, not of the original data.

BTW, assuming pure square waves w/o any harmonic distortion (unrealistic, but interesting to consider), a clipped square wave would look and sound identical to an un-clipped one ;).
theDAWstudio said:
Way too much information...I'll never look at Santa and Moogs the same way again.
Click to expand...
Awww, come on, the idea stimulated you just a little bit I'll bet ;) :D.

Just to give credit where it's due, and to clear things up for newbs to this forum even if it's stating the obvious, that pic is not me. That's actually the pic from the album cover for an old album called "Switched On Santa" - a blatent rip off idea from the "Switched On Bach" album series from Walter Carlos. I just photoshopped out all the original wording on the cover and replaced it with the holiday message.

I'm not really bald, what hair I have is still 98% black, and I can only wish I had a nice big modular Moog like that in front of me :o.

G.
 
masteringhouse said:
Happy Holidays Glen and to all my friends at HR.

May the season be as warm as tube gear and the days be dithered with joy.
Click to expand...
Back at you my friend, and congratulations for being perhaps the first person on the planet to coin the phrase "dithered with joy"; I like it! :D

G.
 
Would dithering still be useful after you already down-converted from 24 to 16 bit?
 
Chili said:
Would dithering still be useful after you already down-converted from 24 to 16 bit?
Click to expand...
Yeah, that's where practically everyone agrees that if you were to use dithering at all that it's most effectively used.

G.
 
Did we switch from red velvet to black velvet?:D
Oops, Merry Christmas!
 
SouthSIDE Glen said:
Yeah, that's where practically everyone agrees that if you were to use dithering at all that it's most effectively used.

G.
Click to expand...

But I mean after it's already a 16 bit file. Normally, you would add dither when it's still a 24 bit file, then down convert. Or do I have that wrong?? Would you normally add dither after it's been converted to a 16 bit file??
 
NYMorningstar said:
Did we switch from red velvet to black velvet?:D
Oops, Merry Christmas!
Click to expand...
Just wait until I go all go all "Blue Velvet" on ya all (a la Dennis Hopper) ;). Merry Christmas, 'Star!
Chili said:
But I mean after it's already a 16 bit file. Normally, you would add dither when it's still a 24 bit file, then down convert. Or do I have that wrong?? Would you normally add dither after it's been converted to a 16 bit file??
Click to expand...
Yeah, dither is added to lat bit(s) of the 16-bit samples after they've been down converted. If you dithered the 24-bit file before chopping off the last 8 bits, the dithering would be chopped off right along with it.

G.
 
SouthSIDE Glen said:
Yeah, dither is added to lat bit(s) of the 16-bit samples after they've been down converted. If you dithered the 24-bit file before chopping off the last 8 bits, the dithering would be chopped off right along with it.

G.
Click to expand...

Doh!! I've been doing it wrong all along. Oh well, I'm sure there's not much difference. :p See, that's why I hang around here, learn something new everyday. (Notice I'm staying out of your "velvet" conversation)
 
I always thought dithering is what drunk people do when walking :confused:
 
You must log in or register to reply here.
Back
Top