Pan law, eq algorithms, etc

[BRIEFCASEMANX]

What do you have your pan law set to in your DAW? Have you noticed a difference in quality with different settings? I have mine set to whatever Cubase SX3 defaults to.

When you EQ something in a digital EQ plugin, but also use input or output gain or both, are all of these adjustments put into the same total algorithm, or is there a separate algorithm for each adjustment? I.E. a gain adjustment will cause tiny errors in a digital file, so if you use multiple gain adjustments in a single plugin, are the errors going to be 3 sets of errors from multiple algorithms being used, or does the plugin look at the entire gain structure and produce one final volume from all the gain changes you have set, thus making rounding errors happen one time instead of a separate rounding error happening with each change made on the plugins interface.

Also if you're using a pan law that has some sort of gain compensation, will the DAW give you a rounding error for that gain change AND the channels gain change, if you've changed that from 0.0db, or will it take both gain changes into account and give you a final gain where you only get rounding errors 1 time, instead of rounding errors for both gain changes.


Sometimes I get paranoid and start freaking the f*** out and my palms get sweaty and I'm afraid to touch anything gain related in my DAW because I don't want a bunch of errors adding up. I hope any of this made any sense outside of my SIKK brain. My NASTY brain. My brain that is loving Surge, and Vault. I'm also starving for Claim Jumpers.

 

[SouthSIDE Glen]

When you EQ something in a digital EQ plugin, but also use input or output gain or both, are all of these adjustments put into the same total algorithm, or is there a separate algorithm for each adjustment? I.E. a gain adjustment will cause tiny errors in a digital file, so if you use multiple gain adjustments in a single plugin, are the errors going to be 3 sets of errors from multiple algorithms being used, or does the plugin look at the entire gain structure and produce one final volume from all the gain changes you have set, thus making rounding errors happen one time instead of a separate rounding error happening with each change made on the plugins interface.

It can be architected either way. Assuming that most plugs are written with an OOP (Object Oriented Programming) architecture, it's likely that - in a perfect world - each control on the plug has it's own code object, but passes it's values on to a central "engine" that actually crunches the numbers.

But this is not a perfect world, and sometimes less idealistic things like the incorporation of legacy code, development deadlines, development team politics, and even the competence of the project architect, can cause deviations from ideal. Unfortunately there is no easy way for us on the outside to tell unless we are developers ourselves with a whole bunch of time on our hands to attempt to de-compile the code or analyze memory data structures and geeky stuff like that. (Or get our hands on the source code ).

But frankly, I wouldn't worry about it. Does it sound good or doesn't it? If it sounds fine, then who cares how they mangle the data? And if it sounds bad, even if it has an ideal software architecture, iit still sounds bad.

G.

 

[danny.guitar]

As Glen said, there's no way to know for sure without looking at the source code or running test WAV data through the plugin and analyzing the output.

Every plugin is made differently. Some programmers take their time and do things right, while others are sloppy, and do things poorly. There's just know way to know.

So you will need to use your ears to judge.

Edit: Also, what is your definition of "errors"? Usually most math is done in 32 bit floating point or even higher (sometimes 64 nowadays), and the only "errors" I can think you are referring to are rounding errors.

 

[BRIEFCASEMANX]

As Glen said, there's no way to know for sure without looking at the source code or running test WAV data through the plugin and analyzing the output.

Every plugin is made differently. Some programmers take their time and do things right, while others are sloppy, and do things poorly. There's just know way to know.

So you will need to use your ears to judge.

Edit: Also, what is your definition of "errors"? Usually most math is done in 32 bit floating point or even higher (sometimes 64 nowadays), and the only "errors" I can think you are referring to are rounding errors.

yeah rounding errors. I guess this could be a small reason why some plugs sound better than others.

 

[kubeek]

yeah rounding errors

I don´t think that rounding errors are an issue here as Cubase and I guess all other programs work and compute with 32bit floating point numbers.
Which makes about 10 orders of precision, and that is damned good precision!

The dynamic range of 32bit numbers is 192dB, which is much better than any recording gear you might have. Even if all the computing rises the noise by say 20dB, it will limit your record to dynamic range of 172dB, which is still great, isn´t it?

 

[Farview]

That is the dynamic range for 32 bit fixed point. 32 bit floating point has a dynamic range around 1500db.

None of it matters. It either sounds good or it doesn't.

 

[BRIEFCASEMANX]

whoa, that's so far below the noise floor that it makes no difference. Could there be any other reason why a track at 0.0db would sound better than a track with it's gain adjusted either way? If not, than awesome, one less thing I have to think about when mixing. Sorry I haven't sent you or southside money, glen. I'm very poor from recording equipment/software. I think I sent you $1 about a year ago. I will send a lot more if/when I stop being poor.

 

[Farview]

You could drive yourself nuts wringing your hands about how the computer is doing something, or you can just listen to the sound and determine if it sounds good or not.

When you break it down, analog gear is far more destructive and introduces far more 'errors' than moving a virtual fader in a DAW.