Satin goes V1.1

[K-Slash]

there is a public beta of v1.1 on page 29

http://www.kvraudio.com/forum/viewtopic ... 87#5503387
rsp

OMG I love you ! Thanks !

[zvenx]

well I assume you mean Urs and Sascha etc
rsp

[Urs]

there is a public beta of v1.1 on page 29

http://www.kvraudio.com/forum/viewtopic ... 87#5503387
rsp

OMG I love you ! Thanks !

Careful... Mac version has a bug... I erased it from the server even..

[Agility]

my friend says it's due to Satin not reporting latency properly, is that true?

No. Satin reports its latency correctly. However, it's up to the host to react properly. Which DAW are you using?
Also, when you duplcicate a track and have an instance on one of these, and when you're flipping the phase on this, you'll notice that even when you press the blue bypass button the two won't exactly cancel out. The overall latency might be perfectly compensated but slight phasing will remain as we're using oversampling (e.g. 8x@44k1/48k, 4x@88/96k) and the filters are minimum-phase but not zero-phase. Of course, the whole processing chain does some heavy phase rotating, which is in the nature of things.

Also with my drums I use the UHX type Compander to beef up drums(magic) by mixing it at like 9 or 3 on the dial and as such you get added noise.... I know you want to keep it natural but is it possible to for a switch to be added to get rid of this noise, to atleast have a choice? I was kinda hoping the new automute feature would affect it but it doesn't.

Reelbus recently added a Compander and theirs doesn't add any noise.

We don't add any noise within the compander section. So, question is where does the noise come from? Is it from the hiss & asperity controls, or is it already existent at the input.

Try making a test:
- load the '-default studio -' preset
- don't apply any processing at the input
- crank up the hiss control to maximum (-40dB)
- select 'uhx I' from the *decoder* list (leave encoder to 'none')
- turn the mix from 0 to 100%. You'll hear the noise becoming quieter. Even more when you select 'uhx II'

When you, instead, choose either uhx type for the encoder only, nothing happens here, as the encoding is placed before the tape noise processing (of course).
BUT: any noise existent at the plugin's inputs gets exxaggerated by any of the encoder systems. This is also natural as all of them work as upward compressors.

And: of course the process of 'noise reduction' with Satin only works with the internal noise sources, as the signal flow is encoder->noise->decoder. The plugin can't filter out any noise from outside, as apposed to 'sigle-ended' noise reduction systems.

If this still leaves you puzzled please provide more info on what you're exactly doing or trying to achieve.

Hi.. I'm on Cubase.

As for the Compander, alike ch00rd says the noise happens mostly at 50, there is no noise from my channels going into Satin and this occurs even with the hiss and asperity at 0... what makes the noise worse is turning the input down, turning the input up higher can decrease and then eliminate this noise though ofcourse you might not necessarily want to drive the into the tape so hard. You get a pumping effect from mixing the compander but this noise gets pumped too, so when for example your kick and snare hit, this noise is pumping up and down to the rhythm.
sometimes I like to have my not driving too much and this then creates more noise and the only way to reduce it is to turn the mix knob on the compander further to full or to 0.. thereby reducing the effect... though I've heard this on the records I'm emulating so I guess it's part of the design, though it'd be nice to have the power of the Compander without this added noise.

Thanks.

[K-Slash]

there is a public beta of v1.1 on page 29

http://www.kvraudio.com/forum/viewtopic ... 87#5503387
rsp

OMG I love you ! Thanks !

Careful... Mac version has a bug... I erased it from the server even..

Thanks for the notice, but I'm on Windows .

[K-Slash]

By the way, if I can suggest offering the ability to enter the values using the keyboard, by double-clicking a knob, it would be perfect !

[wnick]

the update is fine on reaper - no more jumping kobs:) thanks
the lower tape speed is a nice new feature but I miss an updated eq for the low speeds to control the frequency response cause currently it's alittle bit too wobbly. By the way the scaling of the frequency diagram isn't optimal. When missusing the compander section the graph often went out of range.

Mind you, the spectral view is the most accurate at low input levels; the higher the input gain (=distortion) the more crazy the plot becomes. This is usual with all spectrum analyzers using a dirac impulse (like this one). Hence the somewhat twisted figures with low speed, I suppose.

The eq itself is compensated. The pre-emphasis applies automatic gain factors internally (we've stated the factors in the manual).

We might offer selectable analysis method in the future, like log sine sweep which can handle nonlinearities much better than a dirac, at the cost of some ripple.

I know about the low levels for checking frequency response. I've checked it with a level of -20dB.
When I start in studio mode with all knobs at default values except the input knob set ccw, the frequency plot seems to be wrong at the lowest speed setting with activated B type encoder.
I've tried to achieve the old treble boost as used with old cassette decks in the past.

[someone called simon]

I've been looking to find when the intro period discount ends but can't see it mentioned. Somewhere I thought I saw 'at least all of september" but I might be dreaming and it's still not defined. Is it known?

If it's not known, is it known if there will be a warning, or will it just end and I'll kick myself for not saving my coinage fast enough?

[xh3rv]

I've been looking to find when the intro period discount ends but can't see it mentioned. Somewhere I thought I saw 'at least all of september" but I might be dreaming and it's still not defined. Is it known?

If it's not known, is it known if there will be a warning, or will it just end and I'll kick myself for not saving my coinage fast enough?

Oct 7th (from u-he.com front page)

[sascha]

Try this test:
- Select Studio mode;
- Have NO audio input to Satin;
- Bypass Tape (!);
- Input level at minimum;
- Output level at maximum;
- Makeup active;
- Select any of the two uhx modes for Encoder;
- Set Compander Mix around 50%;
- Either leave Decoder mode to 'none' (most apparent), or select one of the two uhx modes.

While the setup is admittedly exaggerated, this results in audible noise, that seems to come only from the Compander section. It's not an issue to me, but I do wonder what causes it in terms of the effect's model.

It's logical that the encoder raises any noise at its input.
We have a bit of 'system noise' in the pre-eq circuits, but very low. If you set input to -20 and output to +12 (32dB gain overall), this noise is ~-100dB RMS.
When you now engage one of the uhx encoders and set mix to 100%, the noise will be at -50dB RMS. This is exactly what the encoder is doing: 2:1 upward compression; it's a straight line. This is exactly what the old dbx systems did.

On the mix issue:
When either uhx decoder is active, level matching has to be performed. These two companders are using a 'rotation point' as threshold, where everything below gets amplified by 2, and everything above reduced by 2. As a result, uhx (as well as dbx formerly) can be used to drive the tape harder. But this resulting level is usually too low to hit the tape directly, so a correction factor gets applied.
In the decoder part, the inverse factor is used before expansion. At 100% mix the levels are 1:1 again. But at anything in-between, the correction factor messes with the signal in the way you've encountered.
So I'd say it's no bug, it's just intrinsic to a system that wasn't designed for being used in a mixed mode.

what makes the noise worse is turning the input down, turning the input up higher can decrease and then eliminate this noise though ofcourse you might not necessarily want to drive the into the tape so hard.

When you lower the input gain and have auto-makeup engaged, you're raising the system noise. That's logical because auto-makeup is compensating for the input volume drop, and it does it at the output. Just like auto-makeup in a compressor.

You get a pumping effect from mixing the compander but this noise gets pumped too, so when for example your kick and snare hit, this noise is pumping up and down to the rhythm.

I recorded a lot of stuff in the 90s using a Tascam TSR-8 (8-track 1/2" machine) which had dbx type-I aboard. What you're hearing is exactly what we all ran into back then: our kick drums and the bass were accompanied by a hiss envelope. It's the system. We learned to live with it back then. At least, it reduced the overall noise floor by ~25dB mostly. These compander systems were designed for narrow-format tapes (and 1/2 inch with 8 tracks is quite narrow), and with every time a system designer cuts the format in half or duplicates the number of tracks on a head, the noise floor doubles (+6dB). So, even the hiss envelope was way better than just turning it off.
You can mimic it by adjusting levels and hiss in a way that the hiss has an RMS of around -60, then activate uhx and you'll hear what I'm talking about

I know about the low levels for checking frequency response. I've checked it with a level of -20dB.

Even that is probably too high for the ultra low-speed mode. Might still work at 3.75, though, but 1.875 is an area where the Satin model runs sort of out-of-specs.
The plot would be better if we used log sines here but still would reveal that levels get crazy below, say, 3ips.

Remember, this is no cassette tape, hence the internal parameters are all set up for a large studio machine (dimensions, physical properties, tape parameters, head gap especially). What we're doing is just speeding down the whole thing. When you'd do that with a real machine (by fiddling with the transport electronics, servos, if possible at all), the result would be equally obscure. 1.875 is actually a lo-fi mode in Satin, due to popular demand after the inital release. We never designed it this way, and I was quite intrigued that it even worked somehow, because I never tried out such low speeds when I developed the model.

[Ch00rD]

Thanks for the elaborate explanation Sascha - I may have to read that two or three times while using Satin.
May I suggest adding a few lines about this stuff in the User Guide as well? Especially the part about the 'rotation point' threshold seems quite essential to a proper understanding of how the Mix parameter works.

[Urs]

Back to test mode:

Satin 1.1 sanity check (2nd atempt):

http://www.u-he.com/downloads/Satin/Satin11Mac.zip
http://www.u-he.com/downloads/Satin/Satin11Win.zip

- revision 1899 (no more fail on AUValidation)
- lower minimum hiss
- hiss & asperity auto mute
- wider range of tape speeds (down to cassette speed)
- improved wow & flutter
- all known/reproducible bugs fixed, including delay screech and Reaper slider flutter

Please check for sanity as we wish to release this tonight

- Urs

[wnick]

I know about the low levels for checking frequency response. I've checked it with a level of -20dB.

Even that is probably too high for the ultra low-speed mode. Might still work at 3.75, though, but 1.875 is an area where the Satin model runs sort of out-of-specs.
The plot would be better if we used log sines here but still would reveal that levels get crazy below, say, 3ips.

Remember, this is no cassette tape, hence the internal parameters are all set up for a large studio machine (dimensions, physical properties, tape parameters, head gap especially). What we're doing is just speeding down the whole thing. When you'd do that with a real machine (by fiddling with the transport electronics, servos, if possible at all), the result would be equally obscure. 1.875 is actually a lo-fi mode in Satin, due to popular demand after the inital release. We never designed it this way, and I was quite intrigued that it even worked somehow, because I never tried out such low speeds when I developed the model.

Ok accepted:)
One idea is left:
What about a tapestop effect and maybe a tapejam you know when the tape is really damaged

[sascha]

One idea is left:
What about a tapestop effect and maybe a tapejam you know when the tape is really damaged

Frankly, such effects were not our major intention with the plugin. We initially wanted a professional high-quality device capable for mastering and studio multitrack work. We wanted to prove that tape is not basically about distortion and other negative artifacts. The 'good' machines were designed to sound transparent and sometimes superiour to (early) digital recorders. With Satin, we wanted to show why this is the case.
We've now enlarged the model to provide more degrees of freedom with 1.1, but we'd have to watch out for not watering the concept of this plugin.

[Urs]

In about an hour we'll have yet another build - because this one still says 1.0.2 for the version of the bundle on Mac.

No worries though, we'll get there...