Comments

PipelineAudio wrote on 5/9/2003, 11:32 AM
this is really getting interesting http://forum.nuendo.com/forum/Forum1/HTML/013343.html

"Apparently Vega’s doesn’t dither back to 24-bit which makes that there is distortion generated.
"

why would he say this? How would he know? It seems to me that truncation way down at that level shouldnt make any difference

Every once in a while I bring up the subject of " the vegas sound ". It always turns into an argument, but there IS a vegas sound, and it is different than other apps. Maybe better, maybe worse, but I think it needs to be inteligently discussed, not just flame fodder

heres a bit from Nika from Sweetwater ( who I dont usually care to hear, but this seems important)
"This randomization will dictate that no two processes will produce the same result! After all, how can they if there is a degree of random variation to their answers?

Therefore, any manufacturers that are producing results that are bit identical are merely truncating their results. As such, they are allowing distortion to be present in their mixes which is a classic sign of bad implementation of digital processing. DITHER MUST BE ADDED AT EVERY BIT REDUCTION STAGE in order to maintain the integrity of digital audio."

this is about comparing the summed files between two apps, and apparently any apps that were bit identical, must not have had dither applied to them, on the way to whatever the target output bit depth was...food for thought

PipelineAudio wrote on 5/9/2003, 2:50 PM
more from Nika and Lynn Fuston

Lynn: "If Digidesign, Logic, DP, Yamaha, SAW, Sonar and Vegas are all doing the same math in a way that yields exactly the same result, then is it your assertion that they all are "broken?"

Nika: "Absolutely and unequivocally. What really concerns me, however, is how many OTHER places in their systems they are using the same types of bad math.

Digital mixers are different from analog mixers in that, because they are all numerically based and work on some numerical analysis provided by Fourier, Shannon, Nyquist and others, they can be 100% accurate and correct. This can only happen, however, if the people writing the algorithms to do this stuff implement the math properly. There are enough ways to screw up the math that I'm raising concern that this one indication of bad math is an indicator of the level of competence of the people that designed the rest of the systems, and if the rest of the system suffers from the same math I am concerned about how the problems can manifest themselves under real world conditions.

People have been complaining about digital mixing for years. Bad math is a perfectly viable explanation for the grounds for their complaints.

"

pretty interesting stuff!

then again maybe just too much geekness
SonyEPM wrote on 5/9/2003, 4:11 PM
Digidesign, Logic, DP, Yamaha, SAW, Sonar and Vegas all suck. Who knew?

PipelineAudio wrote on 5/9/2003, 4:36 PM
LOL

as a fun experiment how about an option to turn on and off dithering at every stage in any bit reduction process in vegas?

what do you think would happen? Anything to write home about?
Rednroll wrote on 5/9/2003, 8:59 PM
"Apparently Vega’s doesn’t dither back to 24-bit which makes that there is distortion generated."

LOL!!! who the hell made that statement? He should get the dumbass oxymoron award. Actually, "dithering" also makes distortion in the technical sense. The definition of distortion is getting something different on the output, than what you originally put in on the input. Infact doing any work on a DAW creates distortion, due to quantization levels in digital audio. You want/don't want dithering, you want an infinite amount of bits? Go back to analog.

PipelineAudio wrote on 5/9/2003, 9:50 PM
gawd I hate to spew any nika-isms, but if it were dithered as we say, the distortion it creates would be UN-correlated with the signal, while a truncated file would have distrotion correlated with the rising and falling of the signal right?

If we take Nika at his word, then the reason we have the accuracy we like ( Vegas' ability to create digital clones, which I have confirmed time and again, to prove it to myself) is WRONG?

But maybe the accuracy at dupeing a stereo file isnt the same type of thing we would want for summing a bunch of files together?

But then again, in any case, except for maybe the very end of a cymbal ring ( well past the noise of the drummers nose hairs) there would be more than enough signal running around not to worry about it anyhow...right?