Sounds weird to me. I would assume that a digital bypass is a true bypass.

Try a test:

1. Import both mixes into a new empty project on two tracks. Make sure they are EXACTLY aligned.

2. Throw one track 180 degrees out of phase.

3. Put the faders at unity. Make sure all settings are identical for both tracks.

4. Set pan at "middle" or "straight up" for both.

5. Play it back.

If they are the same they should cancel each other out, and you'll hear nothing. If you hear something, that's the difference between the two tracks. This experiment will scientifically test your observation, thus ruling out variable we can't account for.

Yeah, even though they're bypassed, they still provide latency to the track just for being there. Some DAW's have a way of making unused plugs muted, so as to not cause latency problems. (also frees up ram etc) Depending on what kind of tracks they're on, it may or may not make a noticeable difference.
JohnTodd's method should clear up any doubt.

I don't think the test can be reliably applied to the wav files I have at present. Here is why. I said the newly restored mix is identical (after making incremental changes) but I should have said it is now audibly indistinguishable. Even if the test indicates a difference, I could not attribute the difference to the bypassed plug ins with absolute certainty. However, I can repeat the whole experiment by exporting two new mixdowns, one with the plugs bypassed, and the other with the plugs deleted. I will try this today and report back. Mind you, regardless of the outcome, I cannot see how the results could nullify my observation that a mix that was clearly not indistinguishable was rendered so merely by adding two missing bypassed plugs, particuarly if this observation re-occurs with brand new mixdowns. But I will leave all doubt and questions aside until I conduct the experiment.

The test reminds me of Bob Carver and his challenge to sonically replicate another companies amplifier, which he did using nothing more than differences. It is an interesting read.

[[url=http://[/URL]="http://www.stereoph…"]The Carver Challenge | Stereophile.com[/]="http://www.stereoph…"]The Carver Challenge | Stereophile.com[/]

I just finished the experiment. I exported the two mixes, one with stereo bus plugs bypassed, and the other with the plugs deleted. I then imported the two to a mastering window, and phase reversed one track. This resulted in a very audible drum track, to the exclusion of everything else. The fact that all other instruments vanished indicates the tracks were synchronized and that the phase reversal worked. I then reverted the phase on that track, and changed the phase on the other track, with identical results.

The residue drum track was mostly high frequency content, yet with some bass. If I were to liken it to an equalizer, it would be something like an extreme smiley face.

I'm on the Droid so hard to answer thoroughly. In my DAWs there are many spots in which to put fx. Some of them send pre fade to other locations. If I turn the fx on the stick itself off but haven't stopped the send portion audio will still be present, just not linearly through the original channel.

Yes, but he's using completed exported mixes. Should cancel that out, too?

If the plug is present and sending audio info it will be included in the final mix.

Furthermore, since the inverted track didn't sum to zero I know the audio is being sent to a bus or track somwhere else, like a verb.

I am not sure I follow what you are saying John. Can you clarify? The bottom line is I have two seemingly identical mixes, except for two plug ins being either bypassed or removed altogether at mixdown. They sound different after the export mixdowns, and different in a way that corroborates with the residue from the phase inversion experiment, that being one track (the one with the plug bypassed) is slightly brighter or more present than the other. To me, the primary reason for the difference must be the two plugs. If you do not think it is the two plugs in question, what cause are you attributing the difference I am hearing, and observing in the phase inverting experiment?

I thought we were under the impression that bypassed plugs do not send information to the mixdown. I was, until this experience.

We are in agreement...there is a difference when there should not be. I thought bypassed plugins did nothing at all. Seems as though they do since the phase thinggie didn't cancel out completely.

Only thing that remains is error-checking. Review everything you've done to make those two exports to rule out some weird mistakes or glitch.

I attribute the difference to the plugs and hem not being completely bypassed. I'll try to make some screenshots later if I get time during baby nap and if my memory works today. I had trouble coming up with glide bolts and agraffes a minute ago speaking with another piano tech so all bets are off :-(

Long story short is that your plugs are still partially active.

Something that would be helpful is if JMM could make screenshot of Cubase with plugs present and another one showing then bypassed.

I suppose I could get screenshots if you really think it's necessary. But the bypass switch is not too difficult to know or describe. Cubase even gives a little text line describing the button's function if you just hover your cursor over the switch. On Cubase, it turns yellow when pressed, thus indicating bypass, and it is comfirmed by hearing the effect vanish from the audio. There appears to be no gradation of partially active or bypassed, and one would surely expect it to be an all or nothing option. Now there is something else to consider. There is another button that deactivates the plug in. I think I will have to run the experiment again to see if deactivating acts the same way as bypassing, or the same way as deleting. In deactivation mode, the description of the plug in remains, but turns grey.

In any case, I am satisfied to have at least established that some kind of minor sonic effect remains when plug-ins are bypassed, as opposed to being removed altogether. When I get a chance, I will run the test with the plugs deactivated instead of bypassed..

The greyed out color is the goal. In my four other DAW's, one can "turn off" the fx rack but all the fx plugins show green-active. If those individual plugins are side chained somewhere else or are sending audio out to a bus themselves then turning off the original channel's fx rack won't stop a thing. Think about it in terms of wiring. Guitar comes into . Guitar goes out to compressor. Compressor goes out to eq. Eq goes out to verb. Verb goes back to but also is sent to an aux. Now it goes down the channel strip itself. By clicking bypass you may simply be cutting off the returning signal of the stick itself and not stopping the sending of the signal down the fx chain in the first place.

This could be a result of whether you have your FX rack set up as pre fade or post fade. I'm not familiar at all with Cubase so I don't know how it defaults or it's particular routing quirks. My point is still that if the FX were bypassed for real they wouldn't affect anything.

I follow your line of reasoning to some extent, but bear in mind that one would fully expect the monitored mix (i.e., what we are listening to) to be a true representation of what will be printed at the mixdown. When I bypass effects, the bypass is confirmed by the ear as sonic changes to the mix.

The difference between the two mixdowns was far more subtle than if the effects had been printed in full in the same way they are heard when not bypassed. It is as if the bypass switch bleeds maybe 5% of the full effect, just enough to be made obvious only through direct A/B comparison, as was the case when I needed to restore the mix to original form and thus did many direct comparisons. Had I not mad this mistake, there is a good chance I may have never noticed. And my mix was simple enough that I have ruled out all other things like effects sends or other routes of bleed, and the two effects were only on the Stereo bus. This remains an unexpected phenomenon.

In other words, "bypassed for real" is apparently not a yes or no decision in Cubase. There appears to be something else called sort of bypassed, depending on how closely you listen. :tongue:

Definitely unexpected. This cannot be replicated with any of my DAW's and I have not assisted any other Cubase user with a similar issue. I'd maybe go check out the Steinberg forums and see if someone else has reported similar.

And of course it could be a faulty code writing of a plugin too. Are these those MDA plugs you downloaded last month?

They are not MDA. They are Bootsy and Kjaerhus plug ins. I like them both, although I am buying a well regarded mastering plug in very shortly, as the Kjaerhus classic limiter cannot compare to commercial mastering plug ins.

I may try to replicate the phenomenon with stock Cubase plug ins, to determine if code writing is a factor. I suppose I could also run the John Todd's difference experiment one plug at a time, to see if it's just one or the other. It's a question of how much time I want to invest in investigating. But then perhaps some other industrious and interested Cubase user might also take up the cause. :smile:

I'd do it if someone wants to buy me a license for Cubase. I have to draw the line at four DAW programs.

Well, the best I can offer is a good lunch, should the occasion ever present itself. :smile: In the meantime, I will update with any new information I discover, if I go deeper on the hunt for a solution to the Great Bypass Caper. :tongue:

Actually, I might be able to do better. I bought two pieces of hardware in the past that came with Cubase LE4, my old Firebox, and my Zoom H4. If I can locate the second disc, I can send it to you, although I forget what hoops you might have to jump through to get it registered.

Is LE4 your current version or have you upgraded?

No upgrades. In fact, there are none for LE4. This version is only supplied to OEM's for inclusion with new equipment. Frankly, since I had a copy when I bought the second such supplied piece of hardware, I did not pay attention to what I did with the disk and packaging, but I will look. Thinking further, I remembered that it does not have to be registered. In the near future, I may purchase Studio or Essential 5, in which case, I could send you the one disk I am sure I can put my hands on. I think this version is very functional, and more than adequate for most purposes. In either case, I'll see what I can do.

 I'm always game for learning new DAW programs.

rofl

Attached files

I have no such inclinations. I think recording is hard enough, with plenty of decisions still to be made beyond the DAW. I knew from the moment I became functionally familiar with Cubase that it would be my last DAW brand. This was solidified when I tried to get my free Abelton Live lite (curtesy of Focusrite) to work, in any small way. I was not only lost, but devoid of any inclination to work my way out of the fog. Cubase is perfectly adequate for me, and I will devote all resources and attention to mastering it, and whatever add ons and plug ins I decide to buy.

Cubase is an excellent program. It just didn't happen to be in the four that I've accumulated over the years.

Granted, that is one of the reasons why I do not need to look further. I'm sure there are other excellent DAW's out there, but the potential return on the time and expense of investigating others (which of course requires much study) was not significant enough, since I knew Cubase was regarded well enough for me. There is also something about ones first exposure to a program. If I had of had first exposure to some other excellent DAW, there is no doubt I would be saying the same things about it instead.

But then people like me can benefit from the knowledge of those like you -who do have multiple platform experience- as surely this provides unique insight and understanding of general DAW function that cannot be attained by those who have only ever used one platform.

This is a total shot in the dark, but I'll add it into this discussion as food for thought...

I have the full version of Cubase 5, so my knowledge is limited to that version and may or may not apply to the version you are running.

Cubase has a feature called "Automatic Delay Compensation" which, as the name implies, automatically adjusts playback to compensate for delays caused by the plugin(s). From what I understand, it is the responsibility of the plugin (those who coded it) to accurately report to the host (Cubase) exactly how many samples of delay are imposed by the plugin so that Cubase can compensate for it. I believe that this delay/compensation routine is still active in "bypass" mode.

If perhaps the plugin was reporting a slightly inaccurate amount of delay then the two resultant files would be slightly out of phase. Perhaps just little enough that only signals of very short wavelength were shifted enough to not cancel. This would explain the remaining high frequency material in your test. I can't explain the bass unless it has to do with multiples of the affected wavelengths... at this point I'm at the extent of my knowledge of these things!

I would suggest trying the test again with other plugins... start with some of the Cubase plugins supplied with the program. At least this will tell you if the problem is specific to your plugin or a bug in Cubase.

Hope this helps,

Jeff

Thanks for the insight. It should help to unsort the root cause of the anomaly.

That would be curious indeed. Every DAW but ProTools has had delay compensation built into it for years. [in fairness PT9 has it in built finally]. That does not mean that the Cubase LE4 couldn't have a flawed implementation of said feature.

Is there a difference between "bypass" and "turning it off" with the little blue button?

All you are doing is describing your ineptitude at proper engineering. Plug-ins that are not being used do not affect any audio in any way. What? Only you are having problems canceling out audio by inverted phase techniques. I frequently play a lot with phase tricks to cancel a lot of things out or to extract instruments or vocals from improperly recorded tracks that included other information that I needed to remove like a music track on top of a vocal track. The isolated music track can be mixed back in, phase inverted, so as to eliminate the redundant music track, that would affect the mix while creating a vocal track without the inadvertently recorded music track on top of the vocal track. So I believe you're subjective observations are flawed by a lack of engineering knowledge. What you are describing is poppycock and your understanding & knowledge of audio engineering is too limited to effect any kind of proper comparison of what you are trying to compare. Your observations would be inferring that somebody's mathematical & programming capabilities are far worse than mine which I don't think is possibleand your knowledge is too limited for you to create these erroneous clueless observations. I'm not saying you're not talented am only saying that you don't have a handle on this stuff like you think you do. Your observations are sort of like saying " I'm pretty certain that my car can go faster the harder I press that pedal on the floor". That would be a knowledgeable observation. Maybe I'm incorrect there? Maybe the comparison should be more like " I've noticed that my car seems to go faster without me in it. I know I'm 100% correct. But I'm not sure?" Sort of like saying a pregnant woman is only " a little bit pregnant" if she's not looking pregnant.

I know I'm not pregnant. I know this because I'm an engineer.
Mx. Remy Ann David

Who are you talking to, Remy?

I think she is talking to me. But I have to respectfully disagree on all counts. When I first noted a problem restoring the mix, and finally achieved indistinguishability through nothing more than the addition of two bypassed plug ins, I then posed the question here, and was provided with a method to reveal any difference. I executed this experiment with brand new mixdowns, and without flaw. I have two mixdowns that are identical, except one has two plugs bypassed, and the other has them deleted, and they sound different. They also reveal a residue drum track through phase reversal of one track. I may be new at recording, but I am certain of my competency as an experimentor, and as an observer of details, regardless of the field in question, or my experience.

And I must confess that I do not appreciate being called inept, so Remy, kindly lay out a precise experimental protocal that would be acceptable to you, and I will duplicate the experiment with your rigor, and concede fault if you can determine my exact error from the earlier experiments.

It is also worth noting that saying my car goes faster without me in it is a true statement, if instead the driver happens to be a horse jockey. The most elemetary physics dictates that there is less inertia for the jockey loaded car to overcome in acceleration. Not everything is as it might appear at first glance, even in our attempts at analogy. :tongue:

I think she was a bit harsh on you...you have clearly said in other threads that you are a newbie. And you are clearly asking for help in this thread.

I can't believe though that there isn't a mistake on your part. Is there a reverb send still active or something? This does seem too weird to me.

Try bypassing and mixing down, and then try turning the plugins off with he little blue circle. I use Cubase 5 so I know what controls you are talking about.

It may well be an unidentified effect, say Addictive Drums own internal reverb bleeding through with the two plugs in bypass mode, but that is diffferent than saying the effect must be my error in controls while executing the experiments. We all seem to agree that bypassed plug-ins should behave at mixdown in the same way as if they were not there, but this does not appear to be the case, at least with my configuration of specific plug-ins and DAW.

I conducted the described experiment without error, and I know what a bypass switch is. I am not that new. :tongue: I also know when bypass switches are active or not active, because I can hear the obvious difference, in conjunction with a change in their colour. The difference in the two exported tracks is not remotely obvious in the same way as when I use bypass switches. It is subtle, and supported by the evidence of a lone residue drum track in phase reversing experiments. If it were my error at mixdown, the residue track from phase experiments would include other instruments. The statement two complex Cubase LE4 mixdowns consisting of multiple tracked instruments and a programmed drum track sound different if they are exported with stereo bus plug-ins bypassed or deleted instead, is a true statement. My job now is to better isolate the unidentifed cause for the effect. I will report back soon, hopefully this evening, with a much more precise understanding, or at least a simplification of methods that others might want to try to replicate.

You are talking about throwing a switch in software. You might hit bypass but if the plug-in program has been inserted and you have " flipped a switch" to bypass doesn't mean that mathematical computations are not included with the bypassed plug-in. It's different in hardware than software. Even on some hardware some equipment bypass switches don't bypass the electronics. Others do. So if you didn't write this program and you think you are turning a switch off in software you're most likely not. And all extra mathematics even in bypass mode will cause a difference that is inconsistent with the original source. And you are hearing the timing and phase difference artifacts as a result of a bypass not truly being a mechanical bypass. If all things were not right in this world MS encoding & decoding wouldn't work properly at all if other timing errors are introduced through other mathematical functions that have to be performed in order to bypass something. I have been relying on 100% cancellation for years. But even in hardware land, if you stick an extra transformer in someplace, you're not going to get 100% cancellation due to the difference in timing versus frequencies. A phase or a timing error will be introduced preventing 100% cancellation when you need things canceled. And you don't know how people are writing these programs. And you can't open up the program to see what's inside conveniently like you can a piece of hardware based equipment. I'm sure you are experiencing an anomaly in your methods. And not necessarily an anomaly in the program or plug-in. Most of these issues are generally and most often operator error. There is so much going on inside these freaking computers. How do you know that a scheduled task is not running in the background or automatically checking your disk for fragmentation or looking for incoming viruses. And since the computer is all based upon clocks as opposed to analog which is based upon well, analog, that moves at a consistent speed, timing errors are only based upon variations of 186,000 mi./s. That's a hell of a lot faster processing speed than any computer we use. Yeah, come to think of it, analog is way faster than digital because digital can't move at the speed of light. So we are still trying to make digital catch-up to analog. Analog has always been the reference standard. We still don't know how to do digital audio properly as DSD by Sony/Philips is way better than PCM by Sony/Philips and we're still not using it much because it's still impractical. And think about this, DSD has been around for over 10 years now. At least Korg is trying to push it to the esoteric stereo recordist. I don't have one but I was one of the first owners of a Sony PCM-F1 back in 1982. Or was that 1983? I didn't care if that had an 11 µs (microseconds) delay between left & right channels in comparison to the professional sounding PCM 1610. It was 1/10 the cost of the PCM 1610. Any response to use Beta videocassettes instead of VHS video cassettes because in Beta machines you could switch off the color dropout compensator. This color dropout compensator could not be bypassed and VHS machines which Sony indicated would cause the Reed Solomon dropout compensation error correction to work much harder. Big deal. My clients never knew how hard my VHS based PCM F1 worked. So, even with 11 µs delays between channels MS cancellation was perceivable to be 100% cancellation. And that still had timing errors that just weren't important enough to hear. If you're hearing that big a change, something is rotten in Denmark. No offense to the Danes. And again I think it's operator error. There's something along the way that's happening that you don't know about. We all go through that sometimes. That's why the video guys are so befuddled by audio guys. They can only deal in precise mathematics to make video work. Whereas audio is like voodoo & black magic and all sorts of improper variables can be introduced that won't keep audio from working the way you want it to. As audio engineers we have way more poetic license technically than do graphics & video folks. That's another reason why we have so many different answers here. Of course there is that possibility that your plug-in hasn't been written well? Given your observations I would presume that even though I'd be fairly convinced that's not the case. I would indeed be upset to think that computer program writers are worse at mathematics than I am. That's just not possible. Your indications, are really indicating that something, is not right in your routing, within the computer. (That's my best William Shatnerese who is just a stones throw away from you)

He's dead Jim... I'm an audio engineer not a doctor!
Mx. Remy Ann David

That makes sense...I'm thinking it might be latency compensation built in to Cubase. Might still be compensating even though the plug in is bypassed.

But Remy, now you have me more perplexed with your line of reasoning than I am with Cubase. Your new remarks -particularly these:

You might hit bypass but if the plug-in program has been inserted and you have " flipped a switch" to bypass doesn't mean that mathematical computations are not included with the bypassed plug-in. It's different in hardware than software. Even on some hardware some equipment bypass switches don't bypass the electronics. Others do. So if you didn't write this program and you think you are turning a switch off in software you're most likely not. And all extra mathematics even in bypass mode will cause a difference that is inconsistent with the original source.

and

But even in hardware land, if you stick an extra transformer in someplace, you're not going to get 100% cancellation due to the difference in timing versus frequencies. A phase or a timing error will be introduced preventing 100% cancellation when you need things canceled. And you don't know how people are writing these programs.

and still other remarks appear to indicate that you believe that some kind of mathematical computation may indeed be included at export with a bypassed plug, and not included with a deleted plug, because the information is obviously no longer there, and that there is the potential it will manifest in some sonic way, and that you actually concur. I cannot square this with your saying at other times that it must be operator error, namely mine. :tongue:

But regardless, I appreciate your remarks, and in spite of my earlier defence, I am going to repeat the experiments with a clear slate, no preconceived notions, and even tighter controls, to try to pinpoint the cause. This may take some time, as I am balancing other non-recording tasks and projects.