Satin: quick mixdown audio demo (tape on / off)

[sascha]

Here comes a quickly-fiddled audio clip of the rock-tune project we had at Musikmesse, remixed it today in Samplitude Pro X.

Mind you, it's only a rough mix, done on headphones, so don't expect a 'real' production... no further fx (except Vandal on the bass), only basic EQ, some tracks treated with Presswerk, and Satin as the multi-track & master tape machine.

Here's without any tape:
http://www.u-he.com/Sascha/_temp/Rock%2 ... %20OFF.mp3

And here with tape (Drums, guitars and backing vox got seperate groups in Satin):
http://www.u-he.com/Sascha/_temp/Rock%2 ... n%20ON.mp3

No further limiting or enhancements whatsoever, RMS around -17dB.

Mixer setup:


[The 'tape OFF' snapshot (as in the 'OFF' audio clip) has all Satins disengaged, except on the tracks where two are serialized; there the first instance is always set up as (tape) delay]

All instances where only used gently. While the difference might be subtle, listen for transients, which are a bit smoothed out (peaks get reduced by up to 3dB), a slightly 'thicker' low end and (all channels running in NAB mode), less harshness in the treble region and probably a bit of a 'gel' effect.

[If anyone's wondering who that singer is: it's Marian Gold from the German synth pop band Alphaville ]

[Simon-Claudius]

Nice! Thanks for this demo. How hard did you hit the tape?

[Dip200]

Sounds good! Very interested in this plugin.

[masterhiggins]

Cool!

I know it's a bit premature for feature requests, but do you plan on ever incorporating an NLS-type saturation on each channel? It would be cool to just have something like Satin across all channels to simulate both mixer channel summing as well as the tape saturation. If not, would you think that the tape emulation would more accurately be placed after the non-linear summing?

Come to think of it, under normal circumstances when recording on tape, the individual tracks are modified at least three times (once on input with all of the different console channels, once when written to tape, and once again when sending back through the console channels while mixing). Has there every been a plugin system/method for accurately modeling this workflow? With Waves NLS, I think the summing all get sent to the bus plugin rather than being passed through each individual channel, so a bounce in place still wouldn't emulate this. Of course I could be wrong.

EDIT: The reason why I asked about the summing feature is that it seems like this plugin was mainly designed to be used during the mixing process across multiple channels to introduce subtle tape saturation, which in turn add up to a noticeable effect. This type of workflow would greatly benefit from an all-encompassing saturation plugin system.

-Sam

[kmonkey]

Cool!

I know it's a bit premature for feature requests, but do you plan on ever incorporating an NLS-type saturation on each channel? It would be cool to just have something like Satin across all channels to simulate both mixer channel summing as well as the tape saturation. If not, would you think that the tape emulation would more accurately be placed after the non-linear summing?

Come to think of it, under normal circumstances when recording on tape, the individual tracks are modified at least three times (once on input with all of the different console channels, once when written to tape, and once again when sending back through the console channels while mixing). Has there every been a plugin system/method for accurately modeling this workflow? With Waves NLS, I think the summing all get sent to the bus plugin rather than being passed through each individual channel, so a bounce in place still wouldn't emulate this. Of course I could be wrong.

EDIT: The reason why I asked about the summing feature is that it seems like this plugin was mainly designed to be used during the mixing process across multiple channels to introduce subtle tape saturation, which in turn add up to a noticeable effect. This type of workflow would greatly benefit from an all-encompassing saturation plugin system.

-Sam

But tape simulation on each channel is already possible with satin. I mean there is some sort of grouping option (if i understood it right). Anyway may i ask why do you think that tape characteristic is same as mixer channel summing characteristic. There are two separate things entirely. I don't think satin would help you there..maybe i am wrong.

[masterhiggins]

But tape simulation on each channel is already possible with satin. I mean there is some sort of grouping option (if i understood it right). Anyway may i ask why do you think that tape characteristic is same as mixer channel summing characteristic. There are two separate things entirely. I don't think satin would help you there..maybe i am wrong.

Sorry. I probably didn't explain my question very well. I'm fully aware that tape saturation and channel summing are two different things altogether. I was referring to what I've observed from both Sascha video demos from Musikmesse and Urs on the forums, Satin wasn't designed to provide substantial amounts of saturation on only a few tracks, but rather very subtle saturation on all tracks. There's even a grouping feature to sum the signals down to groups. I also specifically remember Urs saying that it wasn't ideal to post a single track with saturation on it as an example of before/after as the effect would be too subtle. I'm gathering that the workflow is going to be a bit different than what most people have been using tape saturation plugins for (on occasional tracks, and maybe the master bus).

With this in mind, it appears Satin is going to be doing summing of its own (evidenced by the bussing features), therefore it would be bypassing other methods of summing (such as Waves NLS or Slate VCC). The summing method of Waves NLS, for example, bypasses everything after the channel insert points and routes it directly through to the bus channel of NLS. In this case, they could NOT be sent through and bussed likewise through Satin. It would be one or the other. I suppose they could be run in a parallel fashion, unless the Satin insert point breaks the link to the channel output and sums the signals itself at the master output. I'm wondering how the two could incorporate each other.

Logically, I would think that tape summing and console summing would sound different. Whether the effects would be subtle or unnoticeable is a different matter. Would the sound of the pre-amps, EQ, Faders, etc of a console offer a slightly different non-linear response than a tape machine output (considering the tape/calibration/crosstalk, etc)?

So to sum it up, the question is will Satin allow the possibility of summing through a non-linear summing plugin? Or will it be one or the other? The signal flow design is crucial.

-Sam

[sascha]

Here are two more versions:

All Satin instances running 3dB hotter:

http://www.u-he.com/Sascha/_temp/Rock%2 ... 20+3dB.mp3

Same +3dB level, but all instances set to 'Flat' EQ (was either NAB or IEC15 before). Notice how the highs get muffled and vox & cymbals become overly saturated. Plus, the LF 'oomph' from the NAB curve is missing now:

http://www.u-he.com/Sascha/_temp/Rock%2 ... 20flat.mp3


On the summing thing:
As far as I see it, a major aspect of summing is not only saturation or nonlinear processes per se, but also crosstalk. The more channels you sum up, the more crosstalk alters the perception of 'width'. So, when doing that in a DAW, having all channels interacting is the key. To my knowledge, there is no plugin doing this entirely, at least not on a per-channel basis. It's just not possible with current DAW architectures (I could go into detail here, with timestamping, buffer synchronisation, latency compensation, multithreading, hacking against the play regime of the host.. but I'll spare you the details.). Only thing one could do is have a multichannel plugin (n inputs, 2 outputs), or integrate it directly into a DAW.

Luckily, the 'tape' sound is adding way more colour to a track than pure summing. Sure, a multitrack has its crosstalk, and it's a lot, but it's of course most prominent over neighbouring channels.
In Satin, we implemented crosstalk, lately, but only within a stereo instance. Typically, tape machines can have crosstalk around -60 to -50dB, so it's already quite audible on stereo sources.

[Simon-Claudius]

Sascha,
I have some questions if you don't mind: Have technologies that had already been developed by U-He before you joined the company found their way into Satin (filtering etc.) or is it an entirely new creation?
Also, you've touched on the technical background in one of the Musikmesse videos, but how far did you go to model the behaviour of tape machines? What was the approach (maybe even in comparison to available products)?

Many thanks and all the best,
Simon-Claudius

[EvilDragon]

On the summing thing:
As far as I see it, a major aspect of summing is not only saturation or nonlinear processes per se, but also crosstalk. The more channels you sum up, the more crosstalk alters the perception of 'width'. So, when doing that in a DAW, having all channels interacting is the key. To my knowledge, there is no plugin doing this entirely, at least not on a per-channel basis.

http://sleepytimedsp.com/software/str-bundle/


And free!

[sascha]

I have some questions if you don't mind: Have technologies that had already been developed by U-He before you joined the company found their way into Satin (filtering etc.) or is it an entirely new creation?
Also, you've touched on the technical background in one of the Musikmesse videos, but how far did you go to model the behaviour of tape machines? What was the approach (maybe even in comparison to available products)?

The ingredients are entirely new. Satin is pretty much a text-book approach, started out as a proof-of-concept study. I just wanted to know how the various ingredients of a magnetic recording device interact, how it all works, and most important: what leads to which sound. What happens when you change the biasing? What happens when you change the oscillator frequency? What happens when you fiddle with the gap width? What exactly causes treble loss? (answer: gap loss / gap too wide for a given wavelength; and partly self-erasure). What creates the head-bump effect? (answer: a resonance which builds up at wavelengths equaling the voice-coil dimensions).
So, we've been wading through a deep learning process of our model, for instance, what makes '3D', what makes things 'glue'. While getting the hang on it here, we're pretty happy we didn't have to model a specific device, we could just do our own and explore it all using a wider scope. We could literally craft our own 'components'.

http://sleepytimedsp.com/software/str-bundle/


And free!

Quote: "Simulate the stereo bleeding effect of analog buses with Crosstalk v2"

-> *stereo*

[Doug1978]

^^ Very useful EvilDragon (CrossTalk v2 and Dual Panner particularly so), thanks for the link.

[EvilDragon]

http://sleepytimedsp.com/software/str-bundle/


And free!

Quote: "Simulate the stereo bleeding effect of analog buses with Crosstalk v2"

-> *stereo*

At least something, no?

[sascha]

http://sleepytimedsp.com/software/str-bundle/


And free!

Quote: "Simulate the stereo bleeding effect of analog buses with Crosstalk v2"

-> *stereo*

At least something, no?

Not unusual, many plugins have it. Stereo is simple, you just take either channel, highpass it and add a fraction of it to the opposite one. Of course, 'highpass' can be anything. In circuits, it's mostly due to stray capacitance picked up at various places, so frequency & phase response might be quite twisted. In coils, the flux effects are relatively local, therefore it's not that arbitrary.

[Simon-Claudius]

I have some questions if you don't mind: Have technologies that had already been developed by U-He before you joined the company found their way into Satin (filtering etc.) or is it an entirely new creation?
Also, you've touched on the technical background in one of the Musikmesse videos, but how far did you go to model the behaviour of tape machines? What was the approach (maybe even in comparison to available products)?

The ingredients are entirely new. Satin is pretty much a text-book approach, started out as a proof-of-concept study. I just wanted to know how the various ingredients of a magnetic recording device interact, how it all works, and most important: what leads to which sound. What happens when you change the biasing? What happens when you change the oscillator frequency? What happens when you fiddle with the gap width? What exactly causes treble loss? (answer: gap loss / gap too wide for a given wavelength; and partly self-erasure). What creates the head-bump effect? (answer: a resonance which builds up at wavelengths equaling the voice-coil dimensions).
So, we've been wading through a deep learning process of our model, for instance, what makes '3D', what makes things 'glue'. While getting the hang on it here, we're pretty happy we didn't have to model a specific device, we could just do our own and explore it all using a wider scope. We could literally craft our own 'components'.

Thanks for the insight .
I'm really looking forward to this... It's gonna be a graduation present .

[K-Slash]

Sounds killer !

I really like how it enhances the whole track, without making it too distorded.

Talking about te first "Satin ON" example.

I think the one with +3dB is doing too much.

I'm now wondering how it sounds without Presswerk :p !