Verifying the rendered mix

[Yura]

I had to build mix of dense session, having 22 of tracks and 4 returns, lot of automation, lot of plugins, but only native ones (sonoris, jms and studio), lot of softedges.
built-mix process was about 15 minutes.
In order to check my final mix file I placed it on the empty track, reversed phase of its channel and started to hear the silence.
Of cource, in order to summ this one mixed wave with all summed 22 original tracks, i had to path them on sub-buss #2 assigned to out buss#1, and one mixed file was pathed to sub-bass #3 assigned to buss#1.

i was unpleasantly surprised when in some areas of timeline i heard NOT SILENCE.
researching what can be the reason of this I found the first perpetrator. that was crossfades. Areas with crossfades in the tracks certainly caused something like phase-shifting effect, but i cant see when exactly? - while buildmixing process or while playback..

but x-fades was not only areas having problems while this phase-reverse process verifcation. I heard another weird areas where was not a silence.

Bob , can you give some light what can produce such effect except of x-fades in your engine? seems that in areas where aux sending is - the problems may be

[Bob L]

Yura... Crossfades do not cause any shifting that should eliminate your ability to sum to zero when reverse phase checking the mix.

Try it with straight simple tracks... it works perfectly.

My guess is that as soon as you changed the bus routing... your new mix was no longer the same as the buildmix... no longer a valid check.

You will have to check further...

First... does the buildmix sound fine by itself... is it what you wanted for the finished mix?

I would bet that if you built the same mix 10 times and then compared each of those against the other... they would be sample accurate all 10 times.

Bob L

[R Newlin]

Yura,
This is a little off topic but thought i'd post anyway. My mixes were taking about 15 minutes to do a build mix. I took windows off of background applications, (as suggested here :) )now the same buildmix only takes about 90 seconds.

I think at one time it was recommended to run windows as background applications, to tweak performance, but doesn't apply anymore.

Best Regards,
Rex Newlin

[Yura]

Yura,
This is a little off topic but thought i'd post anyway. My mixes were taking about 15 minutes to do a build mix. I took windows off of background applications, (as suggested here :) )now the same buildmix only takes about 90 seconds.

I think at one time it was recommended to run windows as background applications, to tweak performance, but doesn't apply anymore.

Best Regards,
Rex Newlin

OK, 15 minutes taking for my buildmix is the fastest because the project is big and dense (50 minutes long)

[bit]

If the mix takes 15 minutes to build I guess it doesn't play back in real time without a buffer underrun. And if you start in the middle of the mix while having gates and compressors delays and reverbs and stuff going, it won't totaly phase out. After a little while it more or less should thou. BUT, I know there used to be a timing problem with sequenses followed by sequenses with soft-edges in earlier versions of SS, But Bob fixed that. Are you using the latest version?

Bjorn

[bit]

Sorry Yura - 50 minutes..... I did'nt notice that message before sending the other.

Bjorn

[Yura]

Yura... Crossfades do not cause any shifting that should eliminate your ability to sum to zero when reverse phase checking the mix.

Try it with straight simple tracks... it works perfectly.


I expect it is not absolutely right. at least with this simple test:
make one track having two regions butt-spliced and one long x-fade across them. build mix of this track to the next track. invert phase and start playback within x-fade area. you will not hear sound elimination. but you will get sound elimination if start playback before x-fade area.



My guess is that as soon as you changed the bus routing... your new mix was no longer the same as the buildmix... no longer a valid check.


this is very interesting.
so how buss routing may do change new mix?




You will have to check further...


First... does the buildmix sound fine by itself... is it what you wanted for the finished mix?

I would bet that if you built the same mix 10 times and then compared each of those against the other... they would be sample accurate all 10 times.

Bob L

I did that and i got some errors one time when i builded mix, and at the time switched in other applications. maybe this happened because my swap file was set to 0?

[Cary B. Cornett]

In order to check my final mix file I placed it on the empty track, reversed phase of its channel and started to hear the silence.

i was unpleasantly surprised when in some areas of timeline i heard NOT SILENCE.

but x-fades was not only areas having problems while this phase-reverse process verifcation. I heard another weird areas where was not a silence.



The assumption that a build mix will always come out EXACTLY the same for every pass can be incorrect in certain circumstances. I will give you one example: I would sometimes use an autopan plug in a mix, but would find that the panning "motions" would change from one run-through to the next. Why? Because the modulating "oscillator" was truly free-running, not linked in any way to tempo or timeline (kind of like repeat percussion in the old analog synthesizers, which could never be expected to "match up" with an established click track). Something similar could potentially occur with ANY "modulated" effect.

[Yura]

The assumption that a build mix will always come out EXACTLY the same for every pass can be incorrect in certain circumstances. I will give you one example: I would sometimes use an autopan plug in a mix, but would find that the panning "motions" would change from one run-through to the next. Why? Because the modulating "oscillator" was truly free-running, not linked in any way to tempo or timeline (kind of like repeat percussion in the old analog synthesizers, which could never be expected to "match up" with an established click track). Something similar could potentially occur with ANY "modulated" effect.

This understood and took to the attention before. But I use no any kind of plugin that has such "random" behavior in my project. only EQs, compressors, Ultrafunk reverbs.

[Bob L]

Starting playback in the middle of a softedge will most likely never be sample exact as playing back through the entire softedge... the start calculation is not the same as the in-process running calculation that will ocurr as the data plays through the entire softedge.

I believe this really has no important impact... the end result is whether you like the mix and can recreate it at a later time, regardless of whether it is exactly sample accurate over an entire 50 min project or not.

Let us not forget that analog tape could never even playback the same way twice... back to back. :)

Bob L

[Yura]

But you can look at the situation.
you builded mix and give it to customer on cd.
customer listenes and heares there are some track of very wrong level in some time period.
you open your project and see at that time period everithing is ok as was planned. what you gonna do?
first, you re-build this mix anew. ok, now all is ok, but... there are 20 another builded mixes of another big projects, that may be under doubts of customer from now. what to do? I think there must be some reliable and comprehensive way to do the verification tests.

has anybody any another ideas???

[Cary B. Cornett]

This understood and took to the attention before. But I use no any kind of plugin that has such "random" behavior in my project. only EQs, compressors, Ultrafunk reverbs.

I haven't messed much with the Ultrafunk reverb (although I have it), but I know some reverbs, IIRC, may use modulation tricks (like shifting delay times) to add diffusion and such. I do not know whether the Ultrafunk Reverb has such a capability.

[Carl G.]

But you can look at the situation.
you builded mix and give it to customer on cd.
customer listenes and heares there are some track of very wrong level in some time period.
you open your project and see at that time period everithing is ok as was planned. what you gonna do?
first, you re-build this mix anew. ok, now all is ok, but... there are 20 another builded mixes of another big projects, that may be under doubts of customer from now. what to do? I think there must be some reliable and comprehensive way to do the verification tests.

has anybody any another ideas???
Yura, the Utlrafunk reverb will not allow you to do the kind of testing you're doing because each pass' reverb will be different depending upon where you started playback. If you take out the ultrafunk on the 2nd mixdown then compare it to the one you have (reverse phase)... you'll probably notice the difference I'm talking about.

[Bob L]

Also Yura... the differences you are talking about are not the whole levels being wrong in 20 places... you are seeing very minor discrepencies that cause suttle phasing on a test that could not ever have been done in the history of audio until now anyway... so I feel there is a slight over reaction here possibly. :)

Everything is fine with the mixes and multiple mixes of the same project for all purposes will most likely result in mixes that sound identical to even the golden ears boys.

Now, clients who want to argue about non-existent imaginary issues in their minds... perhaps there is nothing you can do to fix that. :D

Bob L

[Yura]

Yura, the Utlrafunk reverb will not allow you to do the kind of testing you're doing because each pass' reverb will be different depending upon where you started playback. If you take out the ultrafunk on the 2nd mixdown then compare it to the one you have (reverse phase)... you'll probably notice the difference I'm talking about.

Carl, I wonder why you are persisting about Ultrafunk reverb that it will not allow you to do the kind of testing? If you will do this test with Ultrafunk reverb you will hear "out of silence" chunk ONLY at the begining of playback inside the area where UR is on. the length of this chunk correspondes to reverb's "tail" as set. But all rest of hour of timeline where your UR is on you will hear the absolute silence. Just do the test.