Can an expert chime in on why Scope Mixers sound better than Cubase default mixing

[bLiNd]

Its been a minute since ive chimed in here so greetings from las vegas!

I still use the Scope XITE-1 for my daily music needs.

I had a mixdown about 2012 ish. It was for a videogame and I was having a lot of masking happening in the mix. The trumpets would dissappear with the strings coming in and a few other issues. Technically the mix should have been fine. I felt a moment where I heard in my head "use the scope mixer". EUREKA! So i tried!

I didnt do anything but route all the channels into the scope 24/48 mixer and the masking dissapeared. The channel separation was different! I could instantly hear everything in the mix loud and clear.

So my question is why?

I love using the mixer now because of that experience so I do my full mixdowns in Scope

[garyb]

dedicated processing and algorythms.

yes, Scope sounds significantly better to my ears. this hasn't changed.

[bLiNd]

dedicated processing and algorythms.

yes, Scope sounds significantly better to my ears. this hasn't changed.

Yes I had a feeling this was the case. It STILL has this amazingly better sound after all these years of using it (since 2002) even though tech has improved in the native realm. I find myself even reaching for the Masterverb more than my UAD reverbs just because its tone works so much better for my music. So much quality

[LunaMan]

dedicated processing and algorythms.

yes, Scope sounds significantly better to my ears. this hasn't changed.

Yes I had a feeling this was the case. I find myself even reaching for the Masterverb more than my UAD reverbs

Including Capitol Chambers ?

[Gordon Gekko]

So my question is why?

It’s the flux capacitor

[fra77x2]

Better binary

[Gordon Gekko]

Talented engineers

[fra77x2]

It sounds better to cubase default mixing... What is this cubase default mixing?. Just mixing channels?
If you hear scope sound better with just adding channels and nothing more I would say it is fantasy.
It should sound the same

Their is no talent involved when adding numbers

[valis]

1. Pan law may be different between your current Cubase settings and Scope. Many will respond to this point by stating (correctly) that mixing with mono sources tends to give superior results, use stereo only where applicable (overheads, specific presets with wide panning and staging via FX, which of course can still be reproduced in the mixer and disabled in the synth/hardware).

2. Certain plugins are sensitive to gainstaging! Typically compressors, saturators, etc and especially anything that is an 'emulation' will do best if you work within the bounds of what the plugin was designed to operate with in terms of input and output levels. Also while on this subject, the accumulation of actual noise from emulations can increase masking effects, it's generally useful to turn this off where possible (once or twice in a mix where wanted can be ok) as what it's really doing is masking low level artifacts and poor mixing choices.

3. Scope's underpinnings from the Analog Devices API's and math libraries as implemented in Scope itself offer a number of excellent algorithms, one of which is the correct application of TPDF where necessary, which keeps INT quantization errors from accumulating (toggling the LSB with a simple algorithm). Scope is not entirely INT (integer) based, as there are floating point maths in the AD chips as well, but the summing path to my knowledge is 32bit INT, this in combination with judicious use of TPDF gives a very linearized mixing path, which does differ from the software mixing environments you're comparing Scope to.

4. 32bit FLT (and 64bit FLT) became the norm in software mixing because it's easier to implement than integer summing, as the latter requires proper gainstaging from the user and constant bounds checking from the programmer who implements it (to prevent clipping). The tradeoff with floating point is that when you push everything into the red, you don't have to worry as much about clipping if you reduce the master but you also get quantize distortion the further away from 'normal' (where zeroth or median value exists). This is less of an issue than the others mentioned above, but it does increase masking effects due to the acculumation of errors if you use a ton of FX and don't properly gainstage.

[fra77x2]

Exactly as Valis said and a bit more complicated. It is ecstatically complicated and only gifted minds with the ability to grasp extremely complex notions and know myriads of abbrevations can hope to manage. Luckily at no cost.

Valis I think you got a free ticket for Mars for this

[valis]

[bLiNd]

dedicated processing and algorythms.

yes, Scope sounds significantly better to my ears. this hasn't changed.

Yes I had a feeling this was the case. I find myself even reaching for the Masterverb more than my UAD reverbs

Including Capitol Chambers ?

I think it depends on the genre I am working on. Usually for electronic stuff, especially trance, I like using Scope especially

[bLiNd]

It sounds better to cubase default mixing... What is this cubase default mixing?. Just mixing channels?
If you hear scope sound better with just adding channels and nothing more I would say it is fantasy.
It should sound the same

Their is no talent involved when adding numbers

Thats what I thought too, but for some reason it just works better for me. Like i said, all I did was route the channels to the Scope mixer. Masking gone

[bLiNd]

1. Pan law may be different between your current Cubase settings and Scope. Many will respond to this point by stating (correctly) that mixing with mono sources tends to give superior results, use stereo only where applicable (overheads, specific presets with wide panning and staging via FX, which of course can still be reproduced in the mixer and disabled in the synth/hardware).

2. Certain plugins are sensitive to gainstaging! Typically compressors, saturators, etc and especially anything that is an 'emulation' will do best if you work within the bounds of what the plugin was designed to operate with in terms of input and output levels. Also while on this subject, the accumulation of actual noise from emulations can increase masking effects, it's generally useful to turn this off where possible (once or twice in a mix where wanted can be ok) as what it's really doing is masking low level artifacts and poor mixing choices.

3. Scope's underpinnings from the Analog Devices API's and math libraries as implemented in Scope itself offer a number of excellent algorithms, one of which is the correct application of TPDF where necessary, which keeps INT quantization errors from accumulating (toggling the LSB with a simple algorithm). Scope is not entirely INT (integer) based, as there are floating point maths in the AD chips as well, but the summing path to my knowledge is 32bit INT, this in combination with judicious use of TPDF gives a very linearized mixing path, which does differ from the software mixing environments you're comparing Scope to.

4. 32bit FLT (and 64bit FLT) became the norm in software mixing because it's easier to implement than integer summing, as the latter requires proper gainstaging from the user and constant bounds checking from the programmer who implements it (to prevent clipping). The tradeoff with floating point is that when you push everything into the red, you don't have to worry as much about clipping if you reduce the master but you also get quantize distortion the further away from 'normal' (where zeroth or median value exists). This is less of an issue than the others mentioned above, but it does increase masking effects due to the acculumation of errors if you use a ton of FX and don't properly gainstage.

Thank you for this, helps me understand a lot more about what is going on

[fra77x2]

If you just route channels to scope mixer and mix the output with no effects then there should be no difference to adding
channels in your daw. If you think you do then:

a. Simplify your testing setup both for testing your daw mixer and scope mixer
b. Use known waveforms like pulses or dirac impulses in the test. You should get no output when adding a signal with the same signal with inverted phase. Try this both in your daw and with scope. Use a waveform editor able to zoom to individual samples to check waves.
c. Try to eliminate any psychological influence (this is done when using specific test waves and not real audio material)
d. In general there isn't anything that a scope mixer does to an audio track if you just pass it through a channel and scope mixes audio
with sample accuracy exactly as your daw do. Noone would like a bit of his audio tracks altered without applying some proccessing to it.
e. check your daw. Test with a different daw.

Find what makes it better and eliminate it... It should sound the same.

[fra77x2]

Try to load the scope mixer on a single dsp. Perhaps in a multiple channel scenario if some channels are loaded on different dsps they may have diffferent latencies and this could lead to innacuracies on the mix down ( lets say in situations were some channels actually carry parallel processing material) that may appear as improvement on some specific setups but detrimental on others.

Personally I use the scope sdk and simple mix atoms that are always loaded on the same dsp.

So I haven't tested scope's big mixers with multiple asio channels connected really.

But I don't want to confuse you with this. But keep in mind that scope dsps when sending audio between them there is some delay. In the normal scope software (not the sdk) the distribution of atoms on the dsps happens automatically and most of the time it works ok but some times not. Manual assignment of dsp distribution and awareness of what it does gives full control.

I don't know of your expertise but perhaps you should take some time understanding how scope dsp distribution may affect your signals latency and contribute to phase problems.

[valis]

Micro-mixer can lock to a single DSP. The largest mixer will lock channels in groups of 8 stereo pairs if my memory is correct, at least on PCI cards.

I learned to use micro-mixers cascaded to handle bussing and aux groups, and use effects directly in the routing window very early on in my Scope use to avoid phase errors from the larger mixers, and this was later improved (between the phase lock button and Scope allocating things as mentioned above whenever possible) and so my workflow never went back to the larger mixers. I do have SpaceF's more interesting mixers for occasional uses where I want to incorporate the modulators and other features he added, but that's for fun not a core workflow.

What I mentioned about floating point and fixed point is something many users overlook, they tend to keep lowering their master bus in software mixing and pushing individual levels up. Once I learned to gain-stage I was able to get comparable results out of just about every mixer here (hardware, software and Scope/DSP) and so I mix where it makes sense for a signal group or set of bussed objects, and bounce where necessary to make it easier to do final mixing in my DAW.

I do find Scope's TPDF and general architecture is very good for synthesis and Reverb tails, but native has dramatically improved since 2003-2012 when we were first having all of the discussions about this.

Once thing I did NOT mention above, each DAW handles the inline (for Logic it' "track based automation", in Cubase/Nuendo it's "plugin automation" etc) automation differently. For DAWs (like Logic and now Cubase) who use a larger mix (or 'safety') buffer for tracks that are not record armed or permanently monitoring an input, it's generally a multiple of the mix buffer size. For Logic this means automation is calculated in step sizes equal to the 512/1024/2048 samples (small/medium/large mix buffer) size, and Ableton Live was 1024 back when this was measured. Cubase's was one of the largest block sizes, and Samplitude/Sequoia the smallest (64 samples).

In practice the effects of this were negligible on linear fades and simple automation, but curves and extremely fast automation always had problems. For instance if you wanted to chop a drum (kick/snare) using the track automation, you might notice if you're opening the automation to remove the initial part of the drum sound it also truncated the transient you wanted, so you would have to fiddle with teh automation points until it 'sounded' right. On a sine wave or other test tones you can quite clearly hear automated stepping with ramps that are moderate in steepness (not super gentle so the change at the block boundaries is sufficient to hear, but not so fast that you only have a few change points which the ear may not pick up as it's too short).

What this means in practice for mixing is that there's complex automation on the track can actually make the sound of your mix a bit more granular or 'crunchy' from these blocks all being aligned, and so gentler automation of the build in modulators on devices is preferable for heavy manipulation of audio.

Of course in Scope since you have to use MIDI this would be even more pronounced for controls that don't have slew limiting (low pass filtering of the resulting automation move from MIDI input) applied. But this has always been the case for MIDI, and is expected. All the more reason to rely on modular, or for fra77x2 to use the SDK.

Know your tools! And then you can use each for its strengths and minimize its weaknesses.

[nebelfuerst]

What I mentioned about floating point and fixed point is something many users overlook, they tend to keep lowering their master bus in software mixing and pushing individual levels up.

I'm one of "they"
Thank you for this very interesting discussion, which hopefully helps improve my mixing skills !

[Gordon Gekko]

Just gainstage every thing up to +6db
Arf nan, to 11

[valis]