Loving Satin!...but

[xh3rv]

Listen, I don't understand this "hiss" thing. When demoing satin I get no noise from the plug other than that crackle thing.

Is there a step by step to produce this "hiss" everyone keeps talking about?

There's a 'hiss' control in the service panel. Turning it up should get audible. Alternatively one can put an absurd amount of gain after Satin (worth noting IMO: absurd gain staging is discouraged a bit by the metering, input/output). I suppose a bunch of parallel instances of Satin could also stack hiss but, could cancel a lot too?

[mandolarian]

Maybe you got the next version, so the issue is hisstory.

[kvaca]

Listen, I don't understand this "hiss" thing. When demoing satin I get no noise from the plug other than that crackle thing.

Is there a step by step to produce this "hiss" everyone keeps talking about?

There's a 'hiss' control in the service panel. Turning it up should get audible. Alternatively one can put an absurd amount of gain after Satin (worth noting IMO: absurd gain staging is discouraged a bit by the metering, input/output). I suppose a bunch of parallel instances of Satin could also stack hiss but, could cancel a lot too?

no, natural hiss cancels only very rarely...usually hiss adds - another track with same amount of hiss results in more 3-4 db of hiss /RMS/ in mix

[pdxindy]

Across 16 tracks, with 16 compressors adding 6db of makeup gain, with the hiss all the way down on each instance, I'm at the hiss reading -51dBFS

Where did you have asperity set?

[pdxindy]

Listen, I don't understand this "hiss" thing. When demoing satin I get no noise from the plug other than that crackle thing.

Is there a step by step to produce this "hiss" everyone keeps talking about?

I just put 12 instances of Satin in a new project and grouped them so I could tweak them all at once. Turning the hiss all the way down and the asperity also down, the noise is something like -80 so well below audible level

If you start pushing the controls to a bit of extreme settings... saying turning pre-emphasis way up or input or output knobs way up, then with that many instances, you can get the hiss up to where it is quite audible. However, those sorts of settings are outside common range for 12 instances.

In most cases the noise floor will not be an issue. However, I do think it would be worthwhile to have the hiss be even quieter when it is set to minimum so that there is more room for extreme settings with many instances.

[hibidy]

Listen, I don't understand this "hiss" thing. When demoing satin I get no noise from the plug other than that crackle thing.

Is there a step by step to produce this "hiss" everyone keeps talking about?

There's a 'hiss' control in the service panel. Turning it up should get audible. Alternatively one can put an absurd amount of gain after Satin (worth noting IMO: absurd gain staging is discouraged a bit by the metering, input/output). I suppose a bunch of parallel instances of Satin could also stack hiss but, could cancel a lot too?

But that's just it. If you use as directed ( ) it shouldn't be a problem. I can make anything hiss if I overdo it

[hibidy]

Listen, I don't understand this "hiss" thing. When demoing satin I get no noise from the plug other than that crackle thing.

Is there a step by step to produce this "hiss" everyone keeps talking about?

I just put 12 instances of Satin in a new project and grouped them so I could tweak them all at once. Turning the hiss all the way down and the asperity also down, the noise is something like -80 so well below audible level

If you start pushing the controls to a bit of extreme settings... saying turning pre-emphasis way up or input or output knobs way up, then with that many instances, you can get the hiss up to where it is quite audible. However, those sorts of settings are outside common range for 12 instances.

In most cases the noise floor will not be an issue. However, I do think it would be worthwhile to have the hiss be even quieter when it is set to minimum so that there is more room for extreme settings with many instances.

Yeah, that's what I'm saying. I can't make the "hiss" anything that is intolerable unless I really try, and even then, it is pretty tolerable.

I feel like I'm missing something. My ears......not what they used to be but in live or S1 I'd think that if there was some noise floor issue, it would at least show up up on the meters

[djsubject]

Dither me with all this hisstory

[Compyfox]

I don't have a lot of time to read pages of longs posts though so I guess the summary is U-He don't want to remove the noise.

Got it!

They never said they won't. But there just isn't any reason to do so. And this debate (which is too long) clearly focuses on that.

[mellotronaut]

hissterical huh

[Compyfox]

Enough with the hissy fits!

Here's some data I just generated with Cubase 7, SATIN v1.00, most current Slate VTM, and a modified MeldaProduction MNoiseGenerator stream.

Read below.

Code: Select all

Plugins Used:
- MNoise Generator (Pink Noise at -53,3dB) -> MBandPass (HP @2,5kHz 6dB/Oct)
- Slate VTM (Noise Reduction 0dB, Hiss Auto Mute off, Calib @-15dB/default, 16 track, 15ips, FG456 tape)
- SATIN (SE A827 30ips preset)


Technical:
- RMS values measured with bx_meter (k-weighted) and RME DigiCheck (non weighted)
- Mono channels were used
- Pan Rule 0dB
- Measurement Point: 1st Insert on Summing Bus


NOISE FLOOR

1 instance              non Weighted    k-Weighted
MNoiseGenerator (filt)  -68,5dB RMS     -65,4dB RMS
Slate VTM               -65,1dB RMS     -68,2dB RMS
SATIN                   -68,7dB RMS     -71,4dB RMS



8 instances
MNoiseGenerator (filt)  -59,4dB RMS     -56,4dB RMS
Slate VTM               -58,2dB RMS     -55,1dB RMS
SATIN                   -60,9dB RMS     -58,9dB RMS



16 instances
MNoiseGenerator (filt)  -56,4dB RMS     -53,4dB RMS
Slate VTM               -55,2dB RMS     -50,2dB RMS
SATIN                   -58,0dB RMS     -55,9dB RMS



24 instances
MNoiseGenerator (filt)  -54,7dB RMS     -51,6dB RMS
Slate VTM               -53,5dB RMS     -49,8dB RMS
SATIN                   -56,3dB RMS     -54,3dB RMS

What does this data tell us:
1) Noise is adding up if you run it in parallel (confirmed)
2) Even if the tape machines are linked (like Slate VTM), the noise of the whole machine doesn't stay the same - each channel is treated as "own machine" rather than "one system" (currently, no tape machine handles that!)
3) The noise actually gradually somewhat decreases over a certain ammount of tracks (see instance ammount)
4) Since the noise is most prominent in the mids to highs, chances are you roll of a lot of the noise "sound" (harshness) with in series connected console emulations (which add their own noise btw!), or another "to tape" instance (since it's part of the original tape sound)



How to prevent noise:
The most read comment so far is "hiss auto mute". While I understand the reason for it on stopped playback, I don't understand it if there is no signal present while playing (which in turn would make it unrealistic, and mess with the tape/console/pseudo-mastering chain on the summing bus).

To actually reduce the noise in general, "the old days" used noise gates on a per-track basis. Just filtering the noise, then just above the noise floor. Though this was highly impractical for multitrack machines beyond 8 tracks, as you'd waste racks only for reducing the noise.

For this, the Dolby NR modes were introduced. A Compressor/Expander system (or compander), which took care of the signal so that it didn't go into the noise floor, and was "recovered" on playback with the Expander (which created it's own unique sounds). Either that, or simply running the tape hot all the time, and wasting noise gates like there is no tomorrow (again - impractical in hardware form, possible in software form).

Another alternative is to simply turn back the "hiss" (noise) in the plugins. Though this is not always possible, and certain other aspects also interfere (like bias, pre-emphasis noise, etc). Also, certain plugins are not linked in that section.




SUMMARY:
I'm surprised that actually SATIN is considered as "too noisy", while former tape plugin creations (Slate, Nomad Factory, UAD, Waves) are considered not.

Again, I understand a need for "auto mute on stop". But not while playback. Else, if the plugins would be intelligent enough to know "oh wait! We're linked, so no noise in parallel, just one noise floor over all instances", then we wouldn't have this debate.

The sad thing is however, there is no real documentation up until this day, if these machines were measured with "one" channel, and directly at the channel output, or the stereo output. I assume the latter.

So actually, this whole issue can be easily handled for the time being with a globally linked "hiss" knob over all instances (as it is possible with Slate VTM as soon as it realizes "several instances are present, link functions") and an appropriate RMS meter on output (example: summing bus) to find the correct spot for the whole machine. Set up once, use templates in your host.

And the presets need to have the info "this machine uses - xx dB RMS as noise floor, if you use it as multitrack machine rather than individual machine, so please adjust accordingly for realism".

I think I'll do that with my future presets (if the info field will be scrollable).





With that said, I wasted several hours on this - just to find a possible solution where everyone might be happy.

Possible Solution (FRs):

- Introduce a "Noise Control" panel (Service Panel), which includes:
- a "Hiss Auto Mute on Stop" switch
- a "hiss link" switch, so that people can decide to use channel individual hiss (example: Studer A800 channel modules!) or global hiss (example: small scale R2R tape machines)

With these, I think all users will be pleased, as it would offer the best of both worlds. Along with the ability to have a noise floor of > -80dB RMS already, this should be more than enough.

[zvenx]

How to prevent noise:
The most read comment so far is "hiss auto mute". While I understand the reason for it on stopped playback, I don't understand it if there is no signal present while playing (which in turn would make it unrealistic, and mess with the tape/console/pseudo-mastering chain on the summing bus).

I use vst3 when available. Haven't tested it, but I imagine that automatically takes care of your hiss auto mute (no hiss when no audio present on track) vs what I meant (no hiss on stop).
I hope in my paraphrasing I haven't misrepresented your view.
rsp

[Compyfox]

According to the VST3 specs, the plugin should simply "turn off" if there is no signal applied (freeing RAM and CPU power). Yes, that was the case in early VST3 days.

But since VST3 is very buggy in Cubase (for me currently), I can't give any further input on that.


Then again, with my view, I indeed meant using this plugin in general. Not VST3 vs VST2. Which was the orignal main debate as well (not started by me).

[zvenx]

oooh.. I wasn't arguing that there should not be no hiss when no audio is present. I was just saying why that is not an issue for me personally.
rsp

[Ch00rD]

[...]SUMMARY:
I'm surprised that actually SATIN is considered as "too noisy", while former tape plugin creations (Slate, Nomad Factory, UAD, Waves) are considered not.

I could be wrong, but I don't remember seeing anyone claiming that Satin is too noisy while *also* claiming that the others you mention aren't. But the ones I'm somewhat familiar with - the UA Ampex and Studer emulations - both have a switch to disable noise completely. So at least for those cases, this point seems to be completely moot.

Again, I understand a need for "auto mute on stop". But not while playback.

I think it's rather the other way around: when a DAW/host sequencer is not playing, that does not imply that plug-ins should not be processing and outputting sound (e.g. playing a virtual instrument in a host should not require its sequencer to be playing, and neither should a tape effect processing the instrument output); but whenever there is no signal present at Satin's inputs (both during playback and when stopped) there is (subjectively, at least...) an argument to be made for some sort of noise gating / auto-muting.

- a "hiss link" switch, so that people can decide to use channel individual hiss (example: Studer A800 channel modules!) or global hiss (example: small scale R2R tape machines)

Doesn't Satin's parameter lock feature already enable users to opt out of global settings for grouped instances?