Decent Filters?

[renø]

That's not latency.

Latency is per definition, if the (entire) audio output is delayed.

AFAIK latency is a most general word, describing the time between a stimulus and its response. I guess about parameter smoothing this kind of latency won't be audible though (because too short).
Or is there a specific latency definition for audio developpers?

[useruseruser]

That's not latency.

Latency is per definition, if the (entire) audio output is delayed.

AFAIK latency is a most general word, describing the time between a stimulus and its response. I guess about parameter smoothing this kind of latency won't be audible though.
Or is there a specific latency definition for audio developpers?

Yes.
Latency for audio developers means, that the output is delayed in relation to the point of time, the digital system attempts to output the final result.

In other words: additional time offset.

[Alex@PA]

I always liked Prosoniq's NorthPole Resonant Filter (released back in 1998), and still fire it up from time to time. It surely is a classic

Hmm, Northpole doesn't seem to be in the database here at KVR....maybe this is a forgotten gem

Download it here:

http://products.prosoniq.com/cgi-bin/re ... l&refno=27

Cheers!
Alex

[useruseruser]

AFAIK latency is a most general word, describing the time between a stimulus and its response.

I know, every move is based on latency (the latency you mean) finally.

Especially sonic waves (all what we can hear) give us a very good picture of how this works.

[Borogove]

Once again you demonstrate the complete inability to understand that words have more than one meaning, regardless of how closely related the meanings might be.

[useruseruser]

Do we philosophate about "words" or "speech" or "language" (and all the sense behind those things) here

Or do we actually talk about steps in filters and audio processing inside this topic

So what's wrong, if we try to find concrete definitions for our conversation.
Latency has a concreete meaning for audio developers. Which is well defined, btw. You actually should know that.

[Borogove]

Do we philosophate about "words" or "language" and all the sense behind those things here

Or do we talk about steps in filters and audio processing inside this topic

"Latency" has multiple meanings within the topic of audio DSP. Like, oh, just to pick one example completely at random, the time delay between when you adjust a knob and when the sound is affected by the knob change.

Which everyone else in the thread seemed to understand when you came along with your condescension, incidentally.

[useruseruser]

Latency has a concreete meaning for audio developers AND users. Which is well defined, btw. You actually should know that.

Actually I think, you "pick" up what you currently need to proove me wrong.

[useruseruser]

Back to the topic:

One problem regarding that entire theme with audio plugins is, that you can actually click the mouse from minimum to maximum with one touch. And move so fast to other positions, which wouldn't be possible with any hardware.

That's a phenomenon, which don't exist on analog equipment (not even with digital hardware synthesizers).

The Yamaha AN1X. the Waldorf Q or the Virus (are all digital and have allot of knobs) for instance, have a special knob inertia build into them.

Thus it is not possible to tweak them ultra fast for some obvious reasons. Those knobs react, by the way, with a noticeble delay (or if you want: faked "latency").

On our computers we don't have such things, so that it meight be possible to tweak parameters between two samples from minimum to maximum. Resulting in an audible audio click (worst case) or even steps in the signal.

Todays sequencers even give us the possibility to draw in those crazy "unreal" behaviour into automation editors...

Possible workarounds:
There are VST synthesizers and effects, which build a kind of inertia into their visual controls, so that a movement is limited or smoothed. I.e. Virsyn Tera comes to mind or OhmForce plugins and so on.
But the VST GUI standard framework don't support that directly.

But it is not necessary to limit the controls movement that way in my humble opinion, as long the developer implements other things to satify the user with the *possibility* of smooth parameter changes, which don't produce any digital artifarkts.

If MIDI automation is supported, the developer probably *has* to implement a smoothing never the less. Because MIDI has a range of merely 128 values (like mentioned above multiple times).

The Virus, the Q (not that sure), but the AN1X and probably most other professional hardware synthesizers use MIDI EXCESSIVELY. But there are no clicks or steps in the signal. Even the Virus is a very good example, because I never got any digital resonance or filter steps or any other steps with it. This feature is, by the way, *switchable* off or on.

So are VST synthesizers and effects intented to be professional?

So then we probably have the answer.
Fortunately the users, who request such qualities sum up in the last time.

The only way is to implement it, when possible.

[useruseruser]

[edit]

[Borogove]

Latency has a concreete meaning for audio developers AND users. Which is well defined, btw. You actually should know that.

The term "latency" has *more than one* concrete meaning for audio developers and users, as evidenced by the fact that audio developers and users were using "latency" in this very thread in the context of controller response time. Why do you refuse to understand that?

Actually I think, you "pick" up what you currently need to proove me wrong.

I do admit I like to prove you wrong, but it doesn't usually require more than a quick googling, or some back-of-the-envelope math.

[useruseruser]

That dosn't require any comment.

[useruseruser]

Again back to the topic:
(Unlike other persons here, who have nothing to deliver to the topic except bashing arround.)

The other methode mentioned earlier ("let the hosts implement that feature") is in my opinion not exactly the best idea.

The host then had to devide the blocksize of the audio process in such small pieces, that it finally would result in a big performance issue. Thus at each possible automation step (finally mostly all of the time). The division of the block process has then to be moved into an extremely high rate with ultra short buffers to prevent any steps with an audible frequency.

The "crazy jumping arround" of the CPU would then cause a giant performance disaster and any kind of intelligent data pre-caching would be completely impossible that way...

Development excurse: (It makes a significant performance difference, whether a process loops optimized a decent amount of times over a sample buffer or (otherwise) jumps from plugin to plugin, from function to function all the time at nearly each sampleframe or ultra small buffers. In fact you meight get significant higher performance with smart vector processing (not breaking all the time caused by a questionable "host's precision automation" technique at sampleframe basis) ...)

A plugin actually has the possibility to do that (parameter dmoothing) inside its code with a relatively small overhead. The possibility of senseful optimizing and smart pre-caching of sample data remains nearly untouched.

And it can be implemented in a way, which is switchable by the user. If he wants precision automation, he actually can switch it on. If he wants more performance, he can switch it off.

I personally don't like host applications that much, which produce undefined blocksizes all the time to (probably) achieve even that kind of higher automation resolution.
Because it de-optimizes and complicates the internal processing structure and performance of VST audio plugins enormously and meight produce allot of other problems too (as already prooven with FL, where infact more problems with clicks and digital distortion occurs that way - fortunately this can be swithed off in such cases).

[Lunch Money]

My post was fine, j&h... you just didn't seem to understand what I was saying. And of course the developers know more than I do. That's why they're the developers.

[useruseruser]

Yes. And some of those are that smart, that they quickly popped up their reserve member account (to hide their real identity) and then posted even that question ("How do I implement parameter smoothing?") to the developer forum...

Ok, maybe not devs from BetaBugs... Now actually *I* am joking...

So if you are interested in development aspects of the theme, so you meight watch that link (not giving you any guaranty for useful final content) :

http://www.kvraudio.com/forum/viewtopic ... 40#1183840

(btw: There is also our so much loved "thiny god" giving us his highly appreciated knowledge We may all profit from that ... )