ESX24 adsr click

hi,

just strted playing with logic and esx24.

when i turn down the volume envelope adsr parameters down to 0 then i still can hear a click at the beginning of the midi note. i suspect its the beginning of the sample. its a short click somewhere below 50ms.

is that normal in esx24 or am is overlooking something?

logic 9.1.1 osx 10.6.4

thanks in advance!

thank you

strange behavior of an envelope anyway

thank you

strange behavior of an envelope anyway

It's actually pretty common for synths to click when the attack time is set to its minimal value. What do you find strange about it?

This is pretty much normal behavior for most samplers/synths/plugins. There might be some that don't exhibit this behavior, but they're the exceptions.

Curious as to why you're turning down all the controls to zero and expecting different behavior.

And yes, you're hearing the first (very) few milliseconds of the sample, the duration of which reflects the minimum available attack time plus the minimum available decay time of the EXS's volume envelope (ENV2).

BTW, if you were to play a silent sample you wouldn't hear this sound.

Attack clicks are a good thing - it means the envelopes are fast. Synths that don't click have smoothing which means their envelopes aren't as fast. Clicks can often be used to add a bit more aggression in the attack.

If you don't want the click, set the attack to 1 or 2 - it'll be just as fast as your other synths that don't click.

Nothing strange about it at all.

In fact, let's take this one step further with respect to what Beej said...

I just did a little test, and the duration of minimum attack time is about 71 samples, or 1.5 ms. at 48K. That's REALLY fast, highly desirable, and totally normal.

What this implies is that really fast attack transients in samples will not be smeared or masked too much as the volume ramps up from silence to peak level during the attack phase (and with the attack at zero).

The only time a sample can be "dodgy" and produce an unwanted click is if there's a fair amount of DC offset in the sample [edit: or if the sample is truncated at a fairly extreme non-zero crossing]. But that's not usually the case. Anyway, that's all beside the point. Respectfully, I don't think you understand how these things work. As has been confirmed by three of us now, the clicks produced by an envelope set to 0,0,0,0 (ADSR) is normal. But then again, why would you set an envelope that way unless you want to create the sound of a click!!!

I think we have here a situation here where a bunch of us who REALLY understand this stuff have taken the time to explain the whys and wherefores of it all, yet despite all of that someone [ahem] just wants to cry "bug" or "flaw".

...its not the ADSR.. its dodgy samples/sounds...

I agree with that bit though, if there's a click in the attack of a sample, it's the sample, not the ESX24.

Right. And as you pointed out, it's not the fault of EXS (or any sampler).

I'm going to take this one step further (hold on to your hats)...

If the attack of a sampled click (trimmed to the very start of the attack) is played back with a flattened ADSR (all controls at zero) you're going to hear a combination of Logic's ADSR click as well as the attack of the click itself. However, the attack of the sample is going to be amplitude modulated slightly by the rise time of the attack and the fall time (down to zero) of the decay. In other words, the attack and decay times (at zero each) are not instantaneous. The envelope raises the gain of the amplifier from zero to maximum (over a period of approx. 1.5 ms) and falls back to zero at a rate which I haven't measured, but let's just say that's 1.5 ms also. That means that the waveform is being amplitude modulated by a 3 ms amplitude rise and fall in volume which corresponds to a half wavelength at 333 Hz (approximately). So the sound that you'd be hearing isn't "pure" in the sense that envelope opening and closing over a period of 3 ms will "reveal" the pure sound of the sample's attack. Even if it could open and close instantly the mere fact that the volume modulates on/off over the course of 3 ms. colors the sound because of said amplitude modulation. And even if that didn't occur, the mere fact that your speaker cones have to overcome the inertia of that sharp jolt is going to smear the signal.

Let's say your amp envelope was set to zero attack and full-on sustain. The attack of the signal is still going to be amplitude modulated in that 1.5 ms transition from zero to full volume. And while this 1.5 ms is damn quick, the amplitude modulation that occurs can sometimes help to mask DC offset and/or bad zero crossings.

Still, there's little use in setting an amplitude envelope to all zeros and expecting some kind of musically useful result other than the sound of a click.

Now, let's talk about DC offset. If there's enough of it in a sample, the sound of an all-zero'd amplitude envelope will be a much sharper-sounding click than usual. But to 8EBREAKER'S comment about zero crossings, it's not up to Logic to find the zero crossings of the start of the sample. It's up to the sample developer, or it's up to you (if you're rolling your own samples) to ensure that the start of a sample is truncated properly and/or has DC offset removed. See... samplers are dumb playback machines. They don't hold your hand and look for DC offset or badly truncated attacks before they play back the sample. In fact, playback of sound from a sampler isn't that much different from audio track playback! If you recorded something which had a bunch of DC offset, or, you trimmed an audio region so that playback didn't start at a zero (or near-zero) crossing, playback of that track is going to "click!" as soon as the playhead meets the beginning of it.

However, the sampler is a bit more forgiving in the sense that that 1.5 ms of attack can, as explained above, mask the attack of "dodgy" samples sometimes. And of course, if you can't be asked to edit a bad sample, increasing the attack time (and/or changing the slope of the attack) can mask that stuff.

So there it is... if you have dodgy samples, it's up to you to fix 'em! Use the DC Offset Removal function in the sample editor. Or re-edit the sample so that it starts on a zero (or near-zero) crossing.

I have never understood why logics synths click at the start.. its just not right.. to me it sounds like logics synths are not at zero crossings properly, anyway a way to get rid of the clicks in most/sum situations i find is to high cut the frequencies slowly away until the clicking disappears, its not the ADSR.. its dodgy samples/sounds with high frequencies causing the clicks!...

Most of Logic's *synths* don't use samples at all. So if you get attack clicks on the ES1/ES2 or other synths, then that disproves your hypothesis - no samples involved.

Really fast envelopes are a good thing. You can always slow down fast envelopes, but if your envelopes are sluggish, you can't speed them up if you need to.

Really, how hard is it to use an attack setting of 1 or 2, rather than zero?

One of these days you might be programming a hard hitting sound and want some extra grit on the attack, and you'll be grateful for the fast envelopes. Many of the best softsynths are (intentionally) designed this way.

anyway a way to get rid of the clicks in most/sum situations i find is to high cut the frequencies slowly away until the clicking disappears,

That's just a dodgy way of setting the attack time longer.

...its not the ADSR.. its dodgy samples/sounds...

 

I agree with that bit though, if there's a click in the attack of a sample, it's the sample, not the ESX24.

my point is proven when you load up the sub base patch sample in the EXS.. it clicks.. i spent ages trying to get rid of the click using the attack etc but couldn't.. so when i use an EQ and high cut it the click disappears!.. yay!, this happens with quite a few of the EXS patches so it is the samples!.. now when it comes to the ES2 and you have just a sinewave set up it clicks hardcore!???.. i have not yet figured out how to get rid of this click with the ES2, you have to move the attack far to much to get rid of the click.., now i have a hardware synth and have owned a couple in my time and this click crap has never been an issue!.. plus i have used cubase and reason and never had the issue with them either!.. so why oh why does logic!???.

8EBreaker,

The simple answer is that the minimum attack time on those other samples is slower than the EXS's minimum attack time. Solution? Slow down EXS's attack time and/or use the curve parameter to smooth out the attack. What you're complaining about is legitimate (I know the clicks you're talking about) and it's not unique to Logic. It happens in just about any sample playback device ever made where that synth is also designed to have a fast attack time for the amp envelope. But it's avoidable. Increase the attack time.

And... if you're talking about sine wave bass sounds, two things...

1) EXS (and any sampler or rompler, AFAIA) is going to click worse on some notes than others unless that instrument's minimum attack time is slow enough to mask it. If it's fast, like EXS, you have to increase the attack time. Sinewave basses are very 'exposed' in that they have no harmonics to mask the clicks. I could pretty much guarantee that if you were to post an example of the sine wave basses from other platforms you mentioned that they'd have a mushy-sounding attack as compared to EXS.

2) ES2 -- if you have the oscillators set to "free" (and again, you're using sine waves) then there's no way that the synth can be expected to produce consistent attack because the waveform is "live", free-running. When you playback notes is as much a random occurrence as is the phase of the waveform when any note plays. So even if you were playing a one-note bass line, that combination of random occurrences means that some notes will click, some won't. Solutions include:

• setting the attack to be a little longer

• changing the OSC start to "hard" and adjusting the attack time

And if none of that is working, then yes, filter the click on the high end (or notch it out). Had you mentioned sine wave basses previously I wouldn't have said that your approach was wrong because they are particularly problematic. So yes, filtering is a solution, but it is (believe it or not) just as fader8 said -- a "dodgy" way of setting the attack time longer. He can explain.

At the end of the day, use whatever tools or means works for you.

With a superfast attack time, the output waveform is going from 0 (silence) to some positive value immediately. That *is* a click.

If you don't want a click sound, you have to slow down that transient, so the transition from silence to some positive volume is smooth and gradual over some longer period of time (ie, over 50-500 samples, rather than in a single sample value).

Synths that do not click with a minimum attack time, as ski said, simply have slower envelopes - on these, you *cannot* go from zero to a signal in a short enough time to produce a click. Their attack 0 setting is equivalent to a faster synth's attack 1 or 2 setting.

I can't understand why the OP is having such a hard time with this...

8EBreaker,

 

The simple answer is that the minimum attack time on those other samples is slower than the EXS's minimum attack time. Solution? Slow down EXS's attack time and/or use the curve parameter to smooth out the attack. What you're complaining about is legitimate (I know the clicks you're talking about) and it's not unique to Logic. It happens in just about any sample playback device ever made where that synth is also designed to have a fast attack time for the amp envelope. But it's avoidable. Increase the attack time.

 

And... if you're talking about sine wave bass sounds, two things...

 

1) EXS (and any sampler or rompler, AFAIA) is going to click worse on some notes than others unless that instrument's minimum attack time is slow enough to mask it. If it's fast, like EXS, you have to increase the attack time. Sinewave basses are very 'exposed' in that they have no harmonics to mask the clicks. I could pretty much guarantee that if you were to post an example of the sine wave basses from other platforms you mentioned that they'd have a mushy-sounding attack as compared to EXS.

 

2) ES2 -- if you have the oscillators set to "free" (and again, you're using sine waves) then there's no way that the synth can be expected to produce consistent attack because the waveform is "live", free-running. When you playback notes is as much a random occurrence as is the phase of the waveform when any note plays. So even if you were playing a one-note bass line, that combination of random occurrences means that some notes will click, some won't. Solutions include:

 

• setting the attack to be a little longer

• changing the OSC start to "hard" and adjusting the attack time

 

And if none of that is working, then yes, filter the click on the high end (or notch it out). Had you mentioned sine wave basses previously I wouldn't have said that your approach was wrong because they are particularly problematic. So yes, filtering is a solution, but it is (believe it or not) just as fader8 said -- a "dodgy" way of setting the attack time longer. He can explain.

 

At the end of the day, use whatever tools or means works for you.

cheers.. i have read and i'm trying to accept what you say but it just not acceptable to me for logic to be like this.. i just find this so weird, anyway i have just experimented and played around with just the pure tone/sine waves with the ES2 and even when i have the attack at full 10000ms it still clicks..?.. now why???.. the EXS isn't as bad but i still have to have the attack right up there and far to much for me to have a good constant volumed sine, now if i create my own pure tone/sinewave using the test oscillator and bounce it then put it into the EXS there is no click at the start because its zero crossing are fine..?, this is the only way i can figure out as to why the ES2 and EXS click at the start of their own pure tone/sines is because they cant zero cross the start of the sine properly..?.. the envelope should have nothing to do with it.., this just has me so completely baffled.., now my old virus indigo never did this nor does my little microkorg, the sinewaves are a good instant attack at 0ms with no muddy starts and no click at the start... weird!.

There's something else going on.

Is the click coming from Logic, or is it your playback system?

Can you put up a simple example project showing the issue we can take look at?

There's something else going on.

 

Is the click coming from Logic, or is it your playback system?

 

Can you put up a simple example project showing the issue we can take look at?

Its not my system its the logic synths.. look its just so simple to do so why don't you just load up a new project load up the ESX press play on your keys or use the piano roll in logic and you'll notice that the pure tone/sinewave just clicks away, muck around with the attack it still clicks, same with the ES2, plus you shouldn't have to use the attack at all to get rid of a click with a sine thats just weird!, a sine doesn't have a transient for it to click... this to me is the instruments inability to be able to read or create the zero crossing properly and or quick enough for itself.. very annoying.

Its not my system its the logic synths.. look its just so simple to do so why don't you just load up a new project load up the ESX press play on your keys or use the piano roll in logic and you'll notice that the pure tone/sinewave just clicks away, muck around with the attack it still clicks, same with the ES2,

Not here.

With an instant attack, yes I get a click. By the time I raise the attack time to about 1.6ms, the click is gone. And a 1.6ms attack is lightning compared to many hardware synths.

plus you shouldn't have to use the attack at all to get rid of a click with a sine thats just weird!, a sine doesn't have a transient for it to click... this to me is the instruments inability to be able to read or create the zero crossing properly and or quick enough for itself.. very annoying.

No, you are just not getting it.

With synths with free running waveforms, you are not triggering the wave form at the zero crossing point - you are jumping into the waveform at random positions. Which means that at any other position that a zero crossing point (it, 99.9% of the time), the synth is going from silence to some amplitude. If you do that instantly, or with very short times, you will hear a click, because that's what a click is - a very quick jump in amplitude.

The only way to remove the click is to lengthen the time it takes to jump from zero to that target amplitude - which is what the attack time is for.

(The same is true for any amplitude envelope changes - attack, release etc - anything from 1.5ms and faster will click, because that's how we hear. If you have a long attack, but an instant release, you'll also hear clicks on the release.)

When designing a synth, the designer either goes "I do not want to the user to experience clicks, therefore I will limit the envelope stages to go no faster than about 1.5ms:, or the designer goes "I want to give as fast envelopes as possible and let the user decide how to handle clicks". It's a design choice.

Its not my system its the logic synths.. look its just so simple to do so why don't you just load up a new project load up the ESX press play on your keys or use the piano roll in logic and you'll notice that the pure tone/sinewave just clicks away, muck around with the attack it still clicks, same with the ES2,

 

Not here.

 

With an instant attack, yes I get a click. By the time I raise the attack time to about 1.6ms, the click is gone. And a 1.6ms attack is lightning compared to many hardware synths.

 

plus you shouldn't have to use the attack at all to get rid of a click with a sine thats just weird!, a sine doesn't have a transient for it to click... this to me is the instruments inability to be able to read or create the zero crossing properly and or quick enough for itself.. very annoying.

 

No, you are just not getting it.

 

With synths with free running waveforms, you are not triggering the wave form at the zero crossing point - you are jumping into the waveform at random positions. Which means that at any other position that a zero crossing point (it, 99.9% of the time), the synth is going from silence to some amplitude. If you do that instantly, or with very short times, you will hear a click, because that's what a click is - a very quick jump in amplitude.

 

The only way to remove the click is to lengthen the time it takes to jump from zero to that target amplitude - which is what the attack time is for.

 

(The same is true for any amplitude envelope changes - attack, release etc - anything from 1.5ms and faster will click, because that's how we hear. If you have a long attack, but an instant release, you'll also hear clicks on the release.)

 

When designing a synth, the designer either goes "I do not want to the user to experience clicks, therefore I will limit the envelope stages to go no faster than about 1.5ms:, or the designer goes "I want to give as fast envelopes as possible and let the user decide how to handle clicks". It's a design choice.

show some respect here please and do not talk to me like i am and idiot you pretentious butt.., as i keep on saying no matter what i do with the attack it still clicks.. all i want is a good constant sine.., as i keep on saying the only way i can get rid of the click is to use a high cut which is the wrong way to go about it but that is the only way i can stop the click.. now i really really wish i could just put the attack at 1.6ms as you say with your superior arrogance for the click to be gone but it just doesn't get rid of the click at all for me!.. now how many more times shall i repeat myself to you for you to realize theirs a click that the attack doesn't get rid of for me with the ES2 and EXS!.. HHHMMM????.

show some respect here please and do not talk to me like i am and idiot you pretentious butt.., as i keep on saying no matter what i do with the attack it still clicks.. all i want is a good constant sine.., as i keep on saying the only way i can get rid of the click is to use a high cut which is the wrong way to go about it but that is the only way i can stop the click.. now i really really wish i could just put the attack at 1.6ms as you say with your superior arrogance for the click to be gone but it just doesn't get rid of the click at all for me!.. now how many more times shall i repeat myself to you for you to realize theirs a click that the attack doesn't get rid of for me with the ES2 and EXS!.. HHHMMM????.

Which is why I asked for you to provide a sample project that demonstrates the issue, because I am not seeing what you are describing.

show some respect here please and do not talk to me like i am and idiot you pretentious butt.., as i keep on saying no matter what i do with the attack it still clicks.. all i want is a good constant sine.., as i keep on saying the only way i can get rid of the click is to use a high cut which is the wrong way to go about it but that is the only way i can stop the click.. now i really really wish i could just put the attack at 1.6ms as you say with your superior arrogance for the click to be gone but it just doesn't get rid of the click at all for me!.. now how many more times shall i repeat myself to you for you to realize theirs a click that the attack doesn't get rid of for me with the ES2 and EXS!.. HHHMMM????.

 

Which is why I asked for you to provide a sample project that demonstrates the issue, because I am not seeing what you are describing.

i have provided an example which i have explained several times now and will explain to you once again which is load up logic and start a new project, then load an instrument track, then insert the EXS then use the piano roll or your midi keyboard and press the keys to play the tone/sine and it clicks every single time at the start of every single key.. no matter where i put the attack the sine still clicks and the more i go into the attack it defeats the purpose of having the constant volume of the tone i want... now why does it do this?.. the ES2 does it as well.. as i keep on saying the only way to stop this click is with a high cut.. now why?..

I did exactly what you said:

No click.

I did exactly what you said:

No click.

cheers for the help mate.

.. look its just so simple to do so why don't you just load up a new project load up the ESX press play on your keys or use the piano roll in logic and you'll notice that the pure tone/sinewave just clicks away,

Just tried it. I get no clicks using the EXS default sine wave.

.. look its just so simple to do so why don't you just load up a new project load up the ESX press play on your keys or use the piano roll in logic and you'll notice that the pure tone/sinewave just clicks away,

 

Just tried it. I get no clicks using the EXS default sine wave.

well i have had this issue since i first got logic 2 years ago and i am still completely baffled to why this is happening..???... i've spent hours on this again tonight to figure it out and she still clicks..?.. i reckon you guys are deaf so thats why you cant hear the clicks!.. haha.

After sequencing a few low tones using the default EXS sine wave, then bouncing it and observing the waveform, it's apparent that the oscillator retriggers from zero for every note. It is not a free-running oscillator. However, if the release is also zero, a click may occasionally be heard at the end of a note due to the signal being interrupted mid swing.

Try this with your rig and see if you get the same clean attacks with your bounce.