Another thread about summing in scope

[Mike Goodwin]

And as previously determined in this thread between 24-bit and 32-bit it is just different data that is lost. 24-bit favoring the loud portions of audio and 32-bit favoring the quite portions. I hope that I have this all right this time :l

Wrong. 32bit flt can represent every 24bit number AND more.

So for best representation of Scope audio :

1) 32bit int
2) 32bit flt
3) 24bit int
4) 16bit int

Thank you for spelling this out for me. I hope im not driving you crazy. So in my case due to the fact that Live works only with flt. I should be working with 32-bit audio and recording to and from the scope enviroment using the asio2-flt drivers! For god sakes I hope I have this right

[Warp69]

Correct.

[Mike Goodwin]

Correct.

Thank goodness

Go figure, if you want the best results in hosts that use flt. you simply have to use the flt. drivers at the highest bit rate. That seems pretty straight forward now in hind sight. I enjoyed reading as you and Red Muse go over the math. As I said before I did not understand most of it but with a lot of help I now have a much better understanding of what is actually going on. I think it is great that this thread is here and that over time people can refer to it as it is all "on the table". Hopefully it could be used to keep other future debates from taking any more energy at least here in this forum.

[Mike Goodwin]

OK not trying to start a new argument here but I wanted to put any question about the whole plugins that operate with "zero latency" to rest. I wrote Eiosis the makers of Air EQ and got a fantastic email back from them laying it out very clearly. I sent them an email back about a week ago asking if they would mind if I posted some of the information but never heard back. I am going to take the chance and post an except from that email.

First he simply explains that there is an enharent delay when going in and out of any audio interface, ok so we all know that. He then prefaces this statment by saying when dealing strictly with internal processing you can have zero sample latency. And here is the math......

"If you have the most simple plugin on earth, i.e. a wire plugin which has not effect on the sound, output will strictly be equal to input. So, there is absolutely no processing delay.
We have a thing like s(t)=e(t)
s is output
e input
and t is time, representing each sample.

Now, if you have an algorithm which takes the input and process it without the need of a delay, you can have a formula like :
s(t) = F(e(t)) + F'(e(t-1)) + F''(s(t-1)) + etc..

In this formula, F, F' and F'' are some functions, and (t-1) represents the previous sample.

You can see that this algorithm can be processed in real time without any delay, and that we always get something like s(t)=e(t), which is immediate, the other terms in the formula belongs to the past."


So there ya have it. I want to try and keep the non math out of this post as much as possable just to keep things simple. He basicly explains simply some algorithms can be processed with zero delay and others can't because of there formula. He also says that when they are talking about zero sample latency they are always refering to internal latency.

Again this post is in the spirit of sharing factual information.

I would like to finish by saying thank you to Eiosis for getting back to me with a fantastic email and that I sincerely did not intend to offend anyone by sharing it. If there is every any problems about this email please just contact me at mike@michaelgoodwin.net and I will remove the post.

[garyb]

what could possibly be wrong with your post? it's a nice email, and an interesting addition to the thread, but....

dude, he's using semantics on your brain.

vsts are NOT zero latency, EXCEPT within the sequencer, which is what he's telling you. the interface's delay IS the issue.

also, while IN THEORY, it's best to use 32flt(and i'm SURELY NOT telling you not to!), in use, you're not likely to experience any difference in your final product from using the 24bit drivers. just some experience talking, not the word of god....

[Mike Goodwin]

what could possibly be wrong with your post? it's a nice email, and an interesting addition to the thread, but....

dude, he's using semantics on your brain.

vsts are NOT zero latency, EXCEPT within the sequencer, which is what he's telling you. the interface's delay IS the issue.

also, while IN THEORY, it's best to use 32flt(and i'm SURELY NOT telling you not to!), in use, you're not likely to experience any difference in your final product from using the 24bit drivers. just some experience talking, not the word of god....

Good then we agree on all fronts I assure you that he was not trying to mislead me into thinking otherwise. As I already said in my post he in fact explicitly went into that point to make sure that there was no confusion. He was not trying to "use semantics on my brain". Between you and me that type of missunderstanding is all good, but I do not want to misrepresent Eiosis. So again let it be clear Eiosis was not trying to claim that by using there software they would correct the fact that there is currently no way to avoid the latencies produced by our current hardware technologies.

And yes I would be shocked and amased if even when sitting 3 feet from some of the best tweeters in the world I will not be able to hear the difference between 32-bit and 24-bit or even int. vs. flt. and so on.

So again we agree.

[sonolive]

hi mike ,

In this formula, F, F' and F'' are some functions, and (t-1) represents the previous sample.

You can see that this algorithm can be processed in real time without any delay,

He basicly explains simply some algorithms can be processed with zero delay and others can't because of there formula.

hummmm ...
to much confusion ... be carefull ... about your brain

and what if "F" takes 2 seconds to be computed ? with a dualcore 4.7 ghz ? or a AD Sharc ?

i think you are missing the point ...
Digital filters ... (same with analog) "TAKE TIME" to compute ...

in french, we call it "temps de propagation de groupe" (any one could give me here the good "english" term ?), this means :
"how long does it take to a signal (a sample in digital) to cross a system, whatever it is (comp, amp, eq ...)mainly called a filter in digital domain, and system in analog domain ...
and even more "AT WICH FREQUENCY" meaning that a system (filter) don't threat each freaquency equally ...

so ... each system (filter) that "transform" a signal (sample) "takes time to compute" (in digital or analog doamin):
this latency is note expressed in second neitheir in ms but mainly in sample or ns ...
this has mainly no incidence in audio domain (what would be the frequency for a dephase of 5 samples ... try your calculator ...), while we (in fact our ears) work on a small bandwidth ... let's say from 80 hz to 15000 Hz (the one that claim that they can "ear" 40 hz or 18000 hz are generally lyers or mythos ....

so if i could advice you some simple things : (a quote from a post rdmuze, some pages above)

your time is better spent learning some musical theory, arrangements, counterpoint, and mixing...

and i will for myself follow right now his advice ... as i am, myself, a bit poor in counterpoint ! booohhhh , i will work !

cheerz
olive

[garyb]

He was not trying to "use semantics on my brain".

again, i'm not saying anything bad about the developer, except that when they call a plugin "zero latency" and it isn't, that's the misrepresentation. it may not be immoral, or at least it's not considered improper, but it's not true because the important part of the latency issue is ignored.


none of this is really very helpful when it comes to better use of tools or improved recording engineer's techniques...

[Shroomz~>]

while we (in fact our ears) work on a small bandwidth ... let's say from 80 hz to 15000 Hz (the one that claim that they can "ear" 40 hz or 18000 hz are generally lyers or mythos ....

Olive, an accurate recording or sample of a snare can need a bandwidth of around 16kHz as can a recording/sample of hi-hats. Cymbals can require an even higher bandwidth of around 18-19kHz. No myth & no lies !!

Mark

[sonolive]

Cymbals can require an even higher bandwidth of around 18-19kHz. No myth & no lies !!

yep, right

but for myself ... I CAN FEEL THEM ... not ear them ! may be am i too old ?

so speaking, i spent my "engineer" life recording accoustic 'things ..."

cheerz
olive

[Mike Goodwin]

He was not trying to "use semantics on my brain".

again, i'm not saying anything bad about the developer, except that when they call a plugin "zero latency" and it isn't

Well in fact the plug in is zero latency. It is computer hardware that is not

[sonolive]

It is computer hardware that is not

right !

same with analog system !!!

cheerz
olive

[garyb]

He was not trying to "use semantics on my brain".

again, i'm not saying anything bad about the developer, except that when they call a plugin "zero latency" and it isn't

Well in fact the plug in is zero latency. It is computer hardware that is not

correct! that's the point! "zero latency" is meaningless unless the system is zero latency! it is misleading to sell the product based on such operation if the SYSTEM won't operate in that fashion. "zero latency" vsts are strictly a marketing point, having very little to do with their use. there's no latency issue when using almost any vst plugin for mixdown, and still a latency issue using any vst plugin for realtime use, although a 3-6ms latency isn't so long that it should be a show stopper for anyone...

[Mike Goodwin]

Good grief Garyb
At this point I will just asume that you are not intentionaly trying to drive me crazy and move on.

[garyb]

sure, but i don't see the problem.

if you don't like me saying that vsts aren't realtime, i'm sorry. they aren't. no big deal, i'm not calling them useless.

if you don't like me saying that this thread is a bit useless(relating to mixing an excellent sounding recording), i'm sorry. it kinda is. if i were a young engineer or musician, i'd find it confusing, and having limited practical use. that's not a personal attack on anyone of us who participated, though...

[next to nothing]

if a plugin is latency free, its latency free. it doesnt matter what the latency of your system is, if the plugin has 0 time delay from input to output it is latency free. It has nothing to do with the rest of the system. The developer says he has something that can work within a sample. I dont believe him, but that is just common sense, he might have it but time will show.

[garyb]

i don't know...i believe the programmer. i just don't think it matters from a practical viewpoint...

[next to nothing]

...And for the relevancy of the discussion, as long as the guy asking the questions is getting the info he is asking for you cant say its useless The fact that it brings up side questions enlightening his original issue can of course be seen as off topic, but trust me things have gone far worse earlier. Its all relevant issue, and it involves very knowledgable people. What more can a newcomer ask for?

Sorry for going even more OT.

(sorry garry, didnt see your post in between. i totally agree its irrelevant for most of us, and i will still do my summing in Scope. I DO feel i get better results in scope, and i use 24bit ASIO modules:)

[garyb]

[Shroomz~>]

it doesnt matter what the latency of your system is, if the plugin has 0 time delay from input to output it is latency free.

There's some Scope plugs that seem to be latency free.

Sorry Mike, not exactly on topic.