The aliasing thread

[george]

Can you tell me which meter you're using?

Yes, Buzz CPU meter.

[valley]

this post is being read mostly by developers anyway

You'd better hope so.

Hmmm, as one of those 'not developers' reading this thread[1], I can safely say I'm unlikely to buy anything made by William.

OTOH, I'm now seriously considering taking advantage of the z3ta group buy as this thread got me curious enough to redownload the demo (wasn't impressed last time round, but hearing it again, I'm unsure why).


[1] I'll pretend to be interested in the exact SNR performance of LI, but really I just love a good flame fest.

[george]

By the way, this thread has unfortunately unfocused aliasing discussion in order to bring CPU use Don't you think this deserves another dedicated topic, since it is something interesting as well?

[WilliamK]

I just want to say one last time. I'm not using this as a "propaganda" to Wusikstation, as I don't need it. What I was saying is that developes should weight the CPU usage versus audio-quality. Hermite-Interpolation is very good and fast for real-time usage, while any Sinc-Interpolation is very slow. So when comparing programs, it would be more fair to have this in mind. As I see that some companies only got Sinc-Interpolation, and not a "fast but not so good audio quality" option. And IMHO that's not a good thing... that's my point. But somehow, some people take that the wrong way, and we end up AGAIN with a flame-war...

Wk

[WilliamK]

Hmmm, as one of those 'not developers' reading this thread[1], I can safely say I'm unlikely to buy anything made by William.

Is strange on how people get things all wrong. I'm not the devil, I'm a nice guy. So I don't get why people are turning against me... Do you care to tell why? So I can at least defend myself?

Wk

[autloc]

WilliamK - well, it IS a bit provocative to boast the fastest engine on the market, especially when others know that the fundamental algorithms (hermite interpolation, filters, ?) are at least somewhat indebted to musicdsp.org.

[René]

Too bad you fail to see you've been using SFZ as the source for your examples, which features both Linear Interpolation AND realtime-sinc, in a variable number of points.

This gives the user to choose what he/she think it's more important for him/her in both quality and CPU usage terms.

-René

[WilliamK]

WilliamK - well, it IS a bit provocative to boast the fastest engine on the market, especially when others know that the fundamental algorithms (hermite interpolation, filters, ?) are at least somewhat indebted to musicdsp.org.

Sorry, but I don't get what you mean.

Wk

[WilliamK]

this post is being read mostly by developers anyway

You'd better hope so.

Actually that was a joke, sorry I forgot to explain. I said that since I see mostly developers here debating about things. And I'm not the only one getting banged on the head. Is just that I'm the only one to get flamed on the end... hummm...

Wk

[René]

Bram de Jong, where are you. Come here and see the incredible damage you've done to the human kind.

-René

[autloc]

is it not cathartic like a greek tragedy?

no, more of a comedy.

[Scr1pt3r]

George, I'm being nice, as I really don't want to do a flame-war here. I won't change, as it IS true. I tested Jeskola here and is not faster than Wusikstation. Sorry, I guess you just tested the wrong way. Anyway, I won't discuss this anymore, as you are the owner of rights here, you are the master, God of coding... hummmm, but what have you ever coded anyway?

hmm.. unless someone tells me otherwise I dont think george is a coder.
In Discodsp Arguru is the coder while george makes the GUI and web graphics, technical support and etc.

[george]

hmm.. unless someone tells me otherwise I dont think george is a coder.

I can tell you otherwise well, I have done a little stuff, but honestly, nothing relevant enough to be considered useable or the fastest of the market

[arguru]

Bram de Jong, where are you. Come here and see the incredible damage you've done to the human kind.

I think he is moving to Barcelona (Spain) to this thing:

http://www.iua.upf.es/mtg/eng/

congrats dude!.

In Discodsp Arguru is the coder while george makes the GUI and web graphics, technical support and etc.

I preferred to make this public at the release time of Aodix v4, but i'm not longer part of discoDSP (from some time ago). It was after a long time decision and after agreement between George and me.

That doesnt mean that products are death or wont be updated.

In the topic: i dont mean if X samplers offers linear interpolation for realtime or whatever scatological resampling method, if it offers a strict-reconstruction-theorical-perfect-quality for offline rendering (imho, dont care if slow as hell).

Anything else is imho mental masturbation, I would imagine Akai big-boss reading this thread and havin' a (their samplers alias as 'una hiena cociendose en una olla')

Will go for beer. brb.

[george]

Well, I have thought about a conclusion before the CPU spoil arrived:

- Alias is something to be aware of, relevant for developers and end users.
- This full thread will hopefully make any end user aware of yet another feature to look for (or not to) when looking for a sampler that fits its needs.
- Several top notch developers have discussed about implementations and new ways of making anti aliasing, a big cheers to that.
- You can't have a long KvR thread without spoilers but that's life.

I think I'm quite done here, until I manage to get new samplers to be analized, and I hope that takes a long while