Mixing in Scope - Device latencies/phase problems

[Yogimeister]

Hi,

I am diving deeper and deeper into mixing in scope .... And the summing was great - but as I started using "inserts"/devices in scope - I came back full circle as individual tracks sound better - but when summing I start getting horrific phase cancellations and mush ....

(I have gone over the long threads about these issues but didnt see anything too relevant in terms of a solution/workflow ....)


My main issue is the VRBs and FX channels (which are native sends in my DAW and feed out to channels in scope).
Having the VRB(and FX) as a seperate channel means that every time I add a scope device to the main channel - I start getting phase effects between that and the VRB/FX (Most noticeable are stuff like crashes and snares ....)

(I know the lowend is another issue - but it seems like less of a problem - because in the case of send FX - I have (for example) the crashes, lead synth and vocals all feeding the VRB - but if I process each channel later in scope - it seems like a real problem/hassle) to go over and sync all the channels - especially if these settings change every time I load the project ....)



Are there any solutions/methods for this issue??

Also, when trying to emulate a 670 with the 660 comps - i connect two mono 660-M to the L and R channels (or in MS mode via MSslot .... Xenon codec doesnt work for me for some reason ... Awesome archetecture - but the sound seems to "drift" from the stereo center ...??!!) - but I am not sure they are in phase (in any case this is less noticeable than the VRB cancellation artifacts in most cases ...)


Any advice/tips would be appreciated ....
(I am running PCI 5.1 - so no SDK or DSP allocation control I think ...)

Thanx

[astroman]

don't worry - that's just the regular way of a close to realtime system
if you have (or use) latency compensation in your DAW... that's all left to you on the Scope side
it has to - because routing can be arbitrarily complex (which would cause an incedible overhead for 'administration')
some stuff is worth caring, but (depending on source) most doesn't matter
(in analog world and with external fx-processors you also have some non adjusted latency)

you really should not take it too serious - it can drive you nuts to align everything down to the sample level
it won't sound better because it is aligned (in fact many sounds need some 'displacement' to please the ear)

that is a general statement about SFP architecture - your particular 'problem' is slightly different:
whenever you use Asio, there's the buffer latency between the DAW and Scope
depending on your point of monitoring it'll be either too early or too late
(let's just assume a 5ms buffer for simplicity)

it seems your reference is the Scope mixer's output
(as your description suggests that Scope does summing for the master track)
then all native processing from the sequencer arrives 5ms too late in SFP

for compensation you have to delay every(!) Scope channel (or single processor) for these 5ms
as latency is effectively in samples a regular delay doesn't catch up the full range
(the amount of time in milliseconds is for rough tuning, for fine steps you use a sample delay)

the 2448 mixer already has this in the channel master display: the little delay button
once switched it adds 0 to 200 samples of delay
if you don't use that mixer there's a device called PhaseFix for that purpose
it's not that difficult, you have to figured out the values only once, then save them with the project

cheers, Tom

[Yogimeister]

Thanx Tom - though I think my problem is different ... All channels go from mu DAW via ASIO to inputs on a scope mixer (usually stm). Then on the mixer I have devices on each channel/bus - compression on the bass (with 2 kinds of compressors) , then a bus with percussions and a Low Cut and Simple compressor acting as a kick-SC, and a VRB channel which is unaffected.

First problem is that the devices on the bass channel will cause it to be delayed compared to the kick (depending on the devices used, lookahead, etc ...)

The second problem is that since my percussions are also being sent to the reverb in my DAW - i will get nasty phasing problems - like with crashes ....
(This is similar to the problem of parallel processing within scope) - the problem is a result of the original signal being delayed compared to the vrb ...

Now, if I have a lead channel feeding the vrb as well (with no scope devices on it ) then delaying the veb wont help ... And since I cant "negative delay" in scope - the three solutions i can think of is:

1) delay all channels in my DAW (pre send) to compensate for the latency introduced by the scope devices - each channel would have its own delay (ASIO latency doesnt matter here ...)

2). Add "dummy" devices for each channel in scope - set to dry - and hopefully giving the same latency as the "wet/working" devices ...

3). Only use devices on channels when bouncing the channels and use SCOPE only for the master section when mixing (since I still want to use sendFX in my DAW ....)


All 3 seem like a very non-elegant workaround

I was also hoping to SideChain compress in scope to avoid using lots of compressors in my DAW - or sum all my SC channels to a SC-comp group ...

Any insights ?

[astroman]

in fact your problem is slightly different... (I left out this case to keep it simple)
as you apply reverb on the DAW side, this adds extra latency for the full roundtrip
(my examples were just one-way, while the reverb would pass the buffer 2 times)
you send from SFP, the DAW applies reverb and the processed sound goes back into SFP
at least I assume that (for example) Scope Instruments may also receive from this native reverb

except for long lookahead values Scope devices introduce almost no latency
(3-8 samples are common for the more demanding ones iirc)

the most practical approach (imho) is to setup a temporary 'measurement path' parallel to the signal your interested in
that path is set to 'inverted' on the Scope mixer
you solo the instrument and the temporary path
then you insert a delay before the one that's supposed to be earlier
adjust the delay to full cancellation or at least maximum extinction (depends on the process)
at that point the delay will tell you about the difference
(as mentioned you may need a time delay set to 100% wet, 0 feedback... plus a sample delay for fine resolution)

do this test at various interesting points of the routing and you'll quickly get an idea what's going on
forget about the mix for a moment - it would make things over-complicated
one step after the other is easier

if you plan to parallel compress drum sounds you may want to check your Scope devices on that chain, too
(even a few samples will most likely have a noticable influence on the sound in that scenario)

cheers, Tom

[Yogimeister]

Ok - my problem is a bit less sevre (as I feed the verb from my DAW (cubase) - and output it to scope - like the rest of the channels - and I dont feed any scope channels back to the DAW (the only time is when I record the final mix ...)


I mean, I do "send" from scope to my DAW - but only when running scope synths in the composition stage (i later sample them and play them from cubase as audio channels)

I have come to the same conclusion about the "signal measurement" path - but I think I will use the cubase "external fx latency measurer" to calculate the latency times ... I will also check the result by ear ...)


I have tried to "compensate" a bass channel to a kick channel by ear ... The results seem a bit wierd to me - but it was an exercise in adding negative delay in cubase via an insert plugin (voxengo)...
The negative delay i came up with of a channel with SL9000V2 and 2xDAS660M is 0.05 ms ... Which seems a bit "small" .... I will check with a measurement path later on ....
(Although the "phase issues" I heard on the bass were very small to begin with ... Unlike the crash/vrb issue ...)

But is this the "workflow" of mixing through scope in order to avoid phase "smears" and comb "filtering" ?(?) (e.g. Hollow bass and digital sizzles on the crashes ...)

[astroman]

well, from the historic point of view it was pretty straight:
you had a Scope system with the 'nice sound' and a crappy DAW that was best at recording tracks
all audio went to the DAW tracks via Asio, all tracks went back into Scope via Asio
not even need to compensate as it doesn't matter if the 'tape' starts half a second later...
only monitoring of playback and live-tracking needed an adjustment
a lot of Scopers used it that way

but today native audio isn't necessarily crap and you have (want) mixed processing
you just experienced that
as this varies from setup to setup there's no general rule - after all it's down to probing specifically
which can be quite complex (as you noticed yourself)

I'm a lazy guy and I've completely given up on that thing
instead I've separated it into tracking and cut/mixing

for tracking I have 2 functionally identical systems: an iPad with Multitrack-DAW and Scope with VDAT
(the iPad has access to Scope via Adat if required)
both systems track the raw signal plus every fx-channel separately
I don't do midi stuff, it's all more or less a live performance

after recording I have those virtual tapes from either VDAT or Multitrack-DAW
VDAT tapes are moved to the DAW machine by a removable flash disk
the iPad stuiff via WLAN (Multitrack has an internal Webserver)

then I arrange tracks in SAWStudio (on a 2nd Scope system) but use only native processing
(as all main Scope FX have already been captured)
if (extra) Scope processing is needed on an individual track, I send it to Scope via Asio and record output in VDAT,
then make a copy of the region and replace the original in SAW Studio
but that's a rare case - I follow some strategy that evolved by use of the system over time
(I know rather well which fx works best in which context etc)

of course I sacrifice some flexibility that way, but on the other hand the routine delivers fast results
it's mostly about optimizing workflow somehow...
I always try to setup sources as close as possible to the desired results and do only fine tuning in postprocessing
(not intended as a blueprint, but as an example how to get along with a mixed setup)

cheers, Tom

[dante]

What Scope mixer are you using, and on what platform (XITE or PCI) and what DSP's is the mixer using ?

[Yogimeister]

Tom : I dont record - so I only do the second part ... But basically - if you record several drum mics or something like that (figure 8 setup) , dont you get phase issues if you record thru different effects ? (I make electronic music - so its not an organic kick/bass ride - but rather a bunch of "identical samples" flying around in verbs and delays ....
(And P.S. - tracking on an iPad ?? Seriously ??? Wow ... I didnt know it was that powerful ... How many tracks ?? Or just one at a time ?)




Dante: SCOPE PCI 5.1, Stm2448 , not sure if I can check dsp allocations (the mixer is not set to a specific board .... I have two pulsars and one Luna (which I also use for IO) - 15DSPs in total ...



Anyone:
The thing is that even placing a simple lowcut device on a bus seems to alter the phase and the sound of the high freqs (beyond the actual impact of a lowcut) ... Any chance that the STM mixer internal FX (eq and dyn) are phase compensated ?

[Yogimeister]

Well ... I set up the measurement path ... The offset between kick and bass was 0.09 ms - and that actually does sound right (better than the 0.05ms i calibrated by ear - and very similar to when both channels are sent to the same ASIO channel ...

But it does seem like a hassle ... And I also hope that this wont change when I reload the Project (should it ?) .... im guessing latency ight differ regarding which cards the fx chain is scattered on ? Does a multifx device keep the nested fx on the same PCI card ?

[dante]

If I recall from a check.dev test there is a 1 sample delay of STM return from aux.

[jhulk]

only if its set to do it on one dsp

also there is sr compensation some devices in scope dont do it

aqnd if its not set to 48khz then there are samples latency added

i came across this when using bcmodular and doing pitch tuning in that when i set a hz range at 44.1 it was out because of not having sr compensation

since this simon made sr compensation modules

so you need to check that the fx you are using have it other wise you will have phase differences because of samples difference

[djmicron]

try to start with a clean project, no daw, test signals inside scope only, to discover what is going wrong, also if you are using low frequency sounds such as kicks and basses in stereo, with added reverbs and delays, it can easily result in phasing issues, but first you have to check one thing per time until it's clear where the problem comes from.

[Yogimeister]

Is that device working/still available ...

I was not usin auxes - just routing a few channels in the daw that are also sent to a reverb send (which is also outputted to scope ...)

Once i effect any of the dry channels (once the are in scope) they start to phase against the vrb channel ... Same for parallel groups in my daw (if they are sent to a different asio channel)




Regarding the "root cause analysis"/debug - it is definately a problem with phase (as soon as I compensated my bass/kick I suddenly have the booming lowend I should have had ) - and moving channels from the "Kick-SC" bus - back to the normal mix bus (where the reverb channel is also sent) turns the digital mush back into pads (the differences are staggering - even more than the low end - because they are very small delays ....




So far - from my tests yesterday It seems like in the STM there is a 1sample delay for each summing stage (makes some sense) - e.g. -busses come out about 1 sample later than the inputs - and the mix channel comes out 2 samples after the input ....
Regarding fx - I have seen some "strange behaviors" where some effects have much more delay than others - but I havent done in-depth checks on them individually yet ....


@jhulk :
Does scope run at 48k internally ? Not really sure what you meant in what you said (regarding sample delay) - is that like a "sampling offset" or aliasing scenario ?

The things that kindof puzzled me was indeed that the "latency" values were not exactly integers of "samples" (I run at 44.1) ....

Is there anywhere that I can get more info about such "sample offsets/delays" and the "SR" module ? (And how can I tell which devices has it or not ?) . Is it just latency compensation or more of a sample-clock-sync thing ?

[jhulk]

yes scope is sample accurate at 48khz

when set to 44.1 or 96 then there are less or more samples delay

sc modules have sr compensation so that when set to 44.1 or 96 it adds the required sr compensation so its in time but not all devices have it made by other sdk users

and with delays it can really cause issues and why when doing things like pitch or frequency which need to be time accurate setting to work on 1 dsp allows this phase problems not to happen

it tells you in the sdk what the compensation should be for the given frequency rates

a 1 sample delay is nothing but when your daw is already 20-40 samples latency and your adding more things can start to cancel out one way of checking is put it through a mono processor which puts the signals equal to both channels and listen to what happen when switched from mono to stereo if there is massive cancellation then some where the signal is a 180 out of phase to the other channel which is causing total cancellation

with reverb or delay i always put it back through a mixer channel so that i can add eq to remove frequencies that can cancell out the other frequencies

[Yogimeister]

Well, I am not worried about the latency between my DAW and Scope (I am currently running at the HIGHEST latency on this project - as it boosts my CPU resources in the DAW...) because all the channels have that as a "common path" - they all get delayed equally (right ?) and no signals have "round-trips" .

I hope that there are no small differences between individual channels and also the channel-pairs in stereo channels (which might explain the "improved stereo image" in scope
- I am using an STM with phase compensation on.


What seems totally counter-productive is that theoretically - if I add any effect to any channel in scope (especially if it is also post-send ) then I will certainly get a phase change and sound change ....




My grief is that because I am also experimenting with the SCOPE fx - I usually apply them in Solo - and while I only make very small changes with them - which sound good when solo'd - after making a few of those - the whole mix degrades completely - and I didnt catch on to that at first ..... Just made those small changes as kindof a "finishing touch" and exported the mix ... And then had a depressing "morning after" listen - where the mix basically sounds like crap - and I was basically diving in to repare phase issues with more/less plugins - which in turned shifted the phase problems (changing my volume and EQ balances across the whole mix .... In some cases the changes are very severe - maybe even exaggerated later by master section plugins like fat inserts , psyQ and multibands ...)

How do you guys go about mixing in scope (not just summing) on "electronic sounds" ??
(Im guessing that "live"/"organic" performances are a bit less susceptible to these issues ...)

[djmicron]

mixing on the stm is not different than mixing to the real console, i would pay attention to the mid/side part of the mix, and i'd avoid confusion inside the mix.

Organize your mix based on frequency ranges, start from the low end, try to solo the sounds that are on the same spectrum and mix them together, then go ahead by steps, try to use effects such as reverbs and delays as separate channels in wet mode to have more control inside the mix.

I'd suggest to disable phase compensation on the stm, because it can cause phase problems itself.

In the real world too, using effects can cause phase issues, so it's normal if some plugin adds delay, but it's not normal if it's too much.

I remember of some old scope plugins such as the stereo compressors from dadev, that were developed without care of the phase, but the most of the scope plugins are fine here, so i think your issues can be solved with a different mix organization.

I'm just guessing what is going wrong on your side, because i worked on the stm on pci boards for years and i'm satisfied about the results.

Also, i remember on an older pc, there were a pci latency setting inside the bios that caused some unusual latency issue until i discovered it.

[Yogimeister]

Thanks Micron ...

Again, I dont care about the "ULLI latency" (for some reqson I actually get a lower latency when checking than the one stated in the ULLI - but again, that doesnt matter that much ... I just try to align phases a bit when I sample a Scope synth into my DAW for example ....


The main issue (besides the fact that I am now in a "confidence crisis" regarding mixing via Scope) is that I dont see how I can do a proper mixdown without constantly doing phase checks ....

I will post a pic of my setup and an audio example of the following scenario :

I have a lead sound routed from a group channel to an ASIO out (mixer in)- (made up of 3 individual synths playing InTheBox , and each one is sent to different fx and different levels of a vrb send).
I bounce the Lead group (ITB) and the VRB send channel (ITB) - the vrb contains all instruments in the mix - it is not just the send vrb ....

In the STM mixer, the VRB goes into the mix out (no bus) - and I sent the Lead channel to a bus with direct output to a CWcomp (which is side chained to the kick) .... The resulting lead sound is HORRIBLE - sounds high-cut, squashed and a bit distorted .... Last night after understanding that I have major phasing issues - I switched it back to the "noBus" and BAM ! Instant sonic "fullness" ....

(You can basically get an idea of this if you send a channel to a bus but keep the channel "mix active" (green button) - even if there are no fx used - it will phase against itself (thats about 1 sample of latency offset) ....

My plan was to have a SC bus, a low cut bus, a kick bus, a "clean" bus and perhaps another SC/NYC bus .... I mix their direct outs via a second micro/dynamic mixer (i only used 4 busses in this current project)

Now, once I have a vrb bus (or fx bus like delays or chorus or whatever) - routing a channel to a different bus than the vrb or adding any effect - even a simple low cut device - causes phasing ....

I want to eq/compress my bass channel - and there is phasing against the kick and low pads - and the sound (or "eq" of the bass changes)...
i want to compress my vrb ... And its a door to a world of sonic pain - unless I manually compensate the phase after each effect I add (not to mention some fx must have different latencies depending on activated modules ...)

I find it hard to trust that as a comfortable mixing environment ...


Do you guys mix like that??

[djmicron]

i'm sorry, i don't have all your issues, anyway, ulli is used just for asio communication, this is why i suggested you to do tests only inside scope to discover what is going wrong.

Still there is some confusion in your routing description, what is wrong in having 1 sample of latency on a reverb send?

what i mean is that 1 sample is 1/48000 of a second @ 48khz.

If i'm understanding right, you are applying a vst reverb to the leads inside cubase, then the leads are going to scope over asio where you are applying some compression to the dry leads, so you are encountering some delay on the compressed leads, while the wet reverb is coming first, because it's applied inside cubase and it's not processed by other scope effects, so i think you should apply the reverb to the lead after it has been compressed, this is what i would do if this is the case, also, the reverb should contain only the wet signals, not mixed dry/wet ones(not sure if you have bounced wet or a mix).

Then you are having some sidechain wrong results, again, i would apply the compressor before the lead goes into the mixer(or at least as insert inside the channel), i too prefer to get sidechain outside the mixer, but i always apply compression before not after.

As a general rule, do not apply insert effects such as compressors after the mixing stage and do not apply send effects such as reverbs before.

Waiting for some pictures/project examples to understand/help you better.

[garyb]

yep, need plenty of pictures to see what is happening.

there is not even one DAW that can compare with highest-end hardware for sound, and ALL hardware is subject to more serious latencies than in the Scope environment.

[Yogimeister]

Micron, you got it right ...

But I prefer to use a native vrb rather than a scope one (that would solve my verb problem as long as I dont use busses (not to mention busses with fx) or fx after the mixer ...

I cant use a sc before the mixer because I dont want to use 7 compressors for the 7 channels I want to SC.


in any case, yes - that would solve the reverb problem (the vrb track is 100 wet of course) - but it also means that I have to do the same for all send FX (according to this method, i cannot use and "communal" sendfx channels inside cubase (and i usually have some distortion smash, chorus, vrb, delays and the likes as send fx channels)

But that leaves us with phasing of one channel against a different channel ....
Add a compressor to the bass channel and it changes its phase in relation to the kick - which changes its "EQ" .... Add an EQ to compensate and 1) it changes the "EQ" sound of the bass track again (even before you start to apply eq changes)

I found myself trying to heavily boost my kick and bass low end (which was lost due to phasing) - needless to say that the results did not sound so great ....


I measured the delay and it was 4 samples (bass compared to kick) - when I adjusted it - i suddenly was hit by the lowend that was "hidden" (the bass track is actually made of a bass+ sub track - so by itself it has plenty of low end ....)