Hiss and Hum: to let the samples ring, or to not let the samples ring?

[RiffWraith]

http://vi-control.net/forum/viewtopic.php?t=12034

Long story, and I am not looking to dupe the convo here, but I didn't want to open another can or worms over there. Talking about things over there has become difficult enough as it is; I am finding that I am liking that place less and less every time I go there. Anyway...


Well, you wouldn't want to always, but if you don't, you are not getting all of the ambience. So, if you don't want all of the ambience, you don't want the DSR turned all the way up. If you want a dry sound, you turn the DSR down. If you want all of the ambience, you turn DSR up. How much depends on how much you want. Turn DSR all the way up, and likely do not need any extra reverb.


Riff - I think that depends on how the ADSR envelope is implemented with the sample's loop points. Decay and esp. Sustain may be bringing up the tail (or a loop, if there is one) higher than that actual ambience, or even extending it if there is some loop point used. ADSR envelopes can boost the signal disproportionately from the actual sample's natural ambience decay.

Leave sustain at min and I think you'll find the noise is less (still there with the default Decay, but less with no Sustain). More than likely, higher Decay times are also boosting the natural ambience beyond it's true level.

Not that any of this is wrong per ce, but I have seen a similiar mode of thinking from many other people - that the DSR should NOT be brought up. And kdm, I am not picking on you here...:eusa_angel: Of course, we are really talking about staccato and staccato-type samples (like drums, for one thing); this really wouldn't apply to, say, legato sustained strings.

I can think of two scenarios you would not want to bring the DSR up:

1. You want a drier sound

2. If there is a bad inherent problem with the sample(s). A good example here would be extreme noise floor

But, if you want an ambient sound, and there is no problem - or a small one (as in the case with these EW samples) that will not be audible in the end, why would you want the DSR down? Don't you want the natural ambience to ring out? Assuming the room/hall/whatever is good acoustically, and assuming the recording was done properly, letting the natural ambience from the recording ring is alwys going to be better than adding verb later. Why are people opposed to DSR all the way to full?

"ADSR envelopes can boost the signal disproportionately from the actual sample's natural ambience decay. " If so, doesn't that mean there is a problem witht he sample playback engine? I mean, DSR should NOT be doing that - changing the sample or how it's played back. Or am I missing something?

[D]

DSR should NOT be doing that - changing the sample or how it's played back. Or am I missing something?

Sounds like you're missing something. The DSR by it's very nature is restricting the playback of the sample and forcing it to constrain to the artificial parameters set up in the DSR envelope.

[kdm]

I have seen a similiar mode of thinking from many other people - that the DSR should NOT be brought up. And kdm, I am not picking on you here...:eusa_angel:

You are right Riff, there is no technical reason *not* to tweak ADSR however you want - there shouldn't be any significant artifacts from doing so. I decided to leave the thread and details alone - it was getting too defensive rather than simply finding the source of the problem.

Imho, it sounds like an issue with something in post processing that is creating noise on the tails (maybe the ADSR is a factor as well, but since the noise is there before extending DSR, I'm leaning towards post processing or converter noise). The noise I hear in SD2 (same as yours and rgames examples) doesn't sound like analog chain recording noise, unless it's a converter issue, which I would be rather surprised if that were the problem considering the money I am sure went into recording SD2. Hard to say for sure. It isn't a showstopper, but, imho, it warrants some investigation by EW. I would hesitate to use a boosted sample in an exposed setting with that level of noise - in a mix, no problem, but alone I would tweak DSR to cut the tail and add my own reverb/ambience (which I do most often anyway).

Pull up something from SD1 like Taikos Earthquake in Kontakt player, or Kontakt 3 and do the same thing with ADSR - no hiss. The samples in SD2 do sound clearer and more open to me, but there is definitely hiss there that isn't in any SD1 samples that I can find.

D - Riff isn't missing anything. I've tested this here, and there is an issue. DSR does boost gain, but SD2 uses DSR to minimize tails on some samples, so there are longer tails there that shouldn't be a problem with longer releases, etc.

[D]

Issue or not, an envelope, by it's very nature, and regardless of which program is processing it, is a process. As such, it is changing the sample to fit it's own parameters. That is a fact, not an opinion. Carry on. I'm not trying to create an argument. You obviously know more about what he is talking about than I do.

Cheers.

[kdm]

Hey D - you are exactly right, an ADSR is a process - gain and ramping functions mostly. That was my first thought as a potential cause of the hiss, but I'm not sure now, given the way the hiss sounds - it just changes in gain with more decay/sustain, no change in length or spectral content that I can hear.

[Thonex]

I followed that thread from a bird's eye.... didn't read every word... but got the general gist and and even posted on that thread.

Here are my views. First, Nick's stuff has (overall) always been VERY well recorded... and noise and hiss are a fact of all recordings. Period. It is IMPOSSIBLE to record without recording "some" noise. So the question becomes how do you record with the least noise while still using the mic techniques that give you the best sounding results. THAT's the question.

The snippets I heard posted were totally passable for a sample library. I mean people forget how much noise there is in ambient recordings. Go listen to the films score for Schindler's list... there is a fair amount of low rumble and hiss... but it's all part of the experience... and if they noise reduced it... it would have resulted on an unnatural picture of the sound.

When you normalize ppp or p or even mp samples... you'll hear the noise MUCH more.. and if you play those samples back at unnatural levels, it sounds pretty bad because of the elevated noise floor (through normalizing). Compression can have the same effect.

Give me well recorded natural sounding samples over clinically sterile samples any day.... I think Nick accomplishes that. I really believe he got the shaft in that thread.

My opinion.

T

[RiffWraith]

DSR should NOT be doing that - changing the sample or how it's played back. Or am I missing something?

"Sounds like you're missing something. The DSR by it's very nature is restricting the playback of the sample and forcing it to constrain to the artificial parameters set up in the DSR envelope."

No, actually it doesn't sound like I am missing anything.

Are you saying that depending upon how the ADHSR is set, the playback of the sample will be affected? ie - set it to full, and the sonic quality of the sample will differ if the ADHSR is not set to full?

[Thonex]

Are you saying that depending upon how the ADHSR is set, the playback of the sample will be affected? ie - set it to full, and the sonic quality of the sample will differ if the ADHSR is not set to full?

I'll answer that... Yes.

An example (and sorry if I'm not understanding your question):

Let's say you record some very low sound level of fingers tapping on a bass drum. You Normalize it and find that is has a really cool attack... but the noise level after the attack is too much. So you put an ADHSR Envelope on it that follows the general shape of the sound so that after the initial attack, the volume (and therefore hiss) drops down.

This is done quite often... especially for sound design when you want to playback things at unrealistic (but cool) playback levels... and you want to retain the attack but not the hiss afterwords.

Hope this helps.

T

[RiffWraith]

I see what you are saying, but I think we are talking about two different things.

What I am basically saying here is that the sonic quality of the samples is not affected if it runs through the ADSHR envelope, as Bill seemeingly alluded to. You can affect (positively or negatively) the performance of the sample(s) by utilizing the ADSHR controls, but that is not the same as affecting the sonic quality.

Cheers.

[Thonex]

I see what you are saying, but I think we are talking about two different things.

What I am basically saying here is that the sonic quality of the samples is not affected if it runs through the ADSHR envelope, as Bill seemeingly alluded to. You can affect (positively or negatively) the performance of the sample(s) by utilizing the ADSHR controls, but that is not the same as affecting the sonic quality.

Cheers.

Well.... it's a matter of semantics then. You're altering the way it "sounds" so.. in essence you are altering the sonic quality of the palyback of the samples... but not the sample files themselves. Also, ADHSRs can also be used for filters... and other things. But I would agree, that the samples themselves are not altered, just their playback... and what is heard.

Cheers,

T

[kdm]

Give me well recorded natural sounding samples over clinically sterile samples any day.... I think Nick accomplishes that. I really believe he got the shaft in that thread.

My opinion.

T

I agree that noise is a fact of recording, and no doubt I'll take the clarity, depth and image in the newer EW sample libraries over being overly anal about a little recording noise (or chair/stand clicks, thumps from poor sample-start edits, etc ;-). SD2 does sound noticeably better than SD1 in that regard (and SD2 samples aren't edited as short as SD1).

To me it sounds like dither that has been boosted along the way (dithered before normalizing? or some gain added to the dither process, which is possible if using analog dither methods), but it could also be recording noise I guess, though it seems too apparent for a high end recording chain/AD conversion process. My recording chain here has less self-noise and general noise floor. I'm sure EW's chain was better than what I have available.

Either way, some hiss is there in louder samples (mf to f), not just pp or ppp samples. The way I would see it as an issue more than not, is that when listening for top end clarity, this kind of hiss creates a "sandy" sound rather than a glassy sound. Maybe we are just reaching the end of 44.1/48k days. ;-)

Regarding ADSR - it *shouldn't* affect the sonic quality of the samples themselves if only adding gain unless the dsp process is for some reason interpolating samples and creating quantization noise, which can happen esp. with ramp functions (linear, log, etc). But since it will add gain you will hear lower level noise, artifacts, etc more easily. I would expect quantization/interpolation errors to be inaudible though since with 24-bit samples we are talking about -135db or so.