As reported to Steinberg tech support this morning:

I was trying to volume automate a snare drum track using meters to help; noticed different peak levels for the same hits on different playbacks. I was NOT using any plug ins on his track. So I did 2 real time exports of the same material (a soloed snare drum) and found that I got different results! I am talking about peak levels that differ by 4dB! This shouldn't even be possible!

Here are roughly the first 11 seconds of the results of those two exports:

goner snare 1 first 11 seconds.wav
goner snare 2 first 11 seconds.wav

So I cut them off for easy upload, and I didn't bother to cut them at precisely the same spot, but the files should be identical up to that point. They clearly are not. You can measure them, null test them, or hell, just listen - especially to the last 2 hits, they sound quite clearly different.

I used to be of the opinion that different DAWs don't sound different, but it appears as though i may have been seriously mistaken.

Coming to the conclusion that Cubase is not for professional use.

Pretty sure it's the #1 DAW for film/game composers who, apparently, came to a different conclusion.

I don't have it, and I can't comment on your particular alleged bug (not that there's enough info there to make a call if I did most likely), but I find it hard to call folks like Hans Zimmer amateurs

what software doesn't have bugs?

According to whom?

And you are totally ignoring what I've said aside from the one inflammatory remark which I really didn't mean for any one to take personally.

I've used Cubase for 14 years. And I am no less professional than any name you want to drop. I have a record to mix right now, for which I was paid nearly $2000, and I have to do it with an unreliable tool.

However, to address your comment; film and game composers I would guess by and large don't need to use real time rendering, so for them this would not be an issue.

Every poll of professional composers I have seen.

Pretty much Cubase (key selling feature: expression maps), then LogicPro, and then DP.

Like I said, Hans Zimmer uses Cubase, for example (and I could name a bunch of others offhand, but I doubt you'd be familiar with them if you are contentious about this). Kind of hard to think of Hans as an amateur.

Most software has bugs.

But most DAW bugs are annoying at worst. This is a massive, deal breaker sort of bug.

Maybe you don't care if your software changes the sound of your mix erratically and arbitrarily, but I do care that Cubase is doing that to my work.

I've never been a fan of Pro Tools, but unless Steinberg steps up and shows me how to make this stop happening, I will have to switch.

DAW's all sound different due to their different mix engines

Contentious? No, just curious.

Sounds like an automation issue to me. Can you turn OFF the automation and export down to see if it goes away?

That does sound like the most logical place to start troubleshooting.

Yeah, I thought that too, but no, it still happens with automation gone.

Finally I tried creating a new project with just one audio track and then real time exported it twice.

Again the results were NOT identical.

Even if this is the result of some quirk in my system and isn't actually happening to a lot of other people, I still don't think it should even be possible.

My system, if you're curious:

Cubase 7.5.30, Lynx aes16e, driver 2.0 build 019g, Lynx L22, PC, Intel core i7 875K, Asus P7P55-D, 16 gb DDR3 ram, Nvidia Gforce 7300 SE/7200 GS graphics card, UAD 2 Quad, UAD 2 solo, Windows 7 64 bit, up to date

Could you expound on this Doc? Many years ago I loaded a kick sample in Reason 3, Logic X and Reaper and they sounded different to my ears when exported. I didn't try to null them but I did hear slight differences in my "non-rigorous" listening tests. Any info would be highly appreciated.

-FC

Why don't you rollback to 7.5.2? Or even 7.0.7?

Next a step is to try a NON REALTIME mixdown and see if those look the same. That way we can narrow it down to the realtime mixdown as the issue and go from there

One clip is at 88.2/32 bit float and the other is at 44.1/16, so there is a potential for a difference there. Also it seems as if this track was sourced from some sort of softsynth sampler or drum machine. Is that correct? If so, did you render it to an audio track first and then export or did you export directly each time?

Go with another DAW then. Cubase always has a million bugs. It's a complex program. There is no such thing as bugfree software. Best bet is to find an app you like and stay one maybe two versions behind the latest. Make sure you have the updates and hotfixes. That is the best way to minimize impact of bugs. I'm on Cubase 4.5. It's rock solid for me but at the price of not having the latest features.

Yeah, I just noticed this too. There's too much inconsistency in the data to make any kind of meaningful diagnosis.

Didnt catch that. 2nd question is a really good one as well

already did that (sorry, thought i mentioned it) non real time works fine.

The differences are HUGE - way bigger than the differences between 44.1 and 88.2. Meant to convert both, didn't even realize I could get the 88.2 to upload. That was the original sample rate.

It's a real f#cking drum, just manually gated. Can't you hear that?

bullsh#t.

it's a serious bug, not user error. We have 2 real time renders of the same audio with results that vary in audible ways. The 9th hit actually has a difference in peak level of 4dB! Sample rate conversion wouldn't cause that.

I would have guessed it was a sample. I've never heard a snare played as part of a kit where the track didn't have at least a little of the other components in it, gated or not. Interesting that you could do it.

And FWIW, in WAV #1, you can clearly hear the level jump right after hit #8, well before hit 9. I sure wouldn't my DAW arbitrarily doing that -- hope you get it solved.

Don't kid yourself, all the big movie projects end up in ProTools regardless of what DAW gets used to start from. Someone like Zimmer wouldn't care about this kind of problem because the mix would happen in ProTools.

Just to understand , You have a single shot snared that you retriggered and some of the hits are louder by a bit . I am curious if you just took the first hit and lopped it and played it back if it would do the same thing . If you retriggered it - could it be the trigger program causing the initial rendering problem ?

Dude, there's no reason to get all testy and start swearing at people. Especially the people who are actually trying to figure out your issue for you.

What i hear are varying snare hits with varying cymbal bleed which fade to nothing in between the hits. Some samplers round-robin samples so that even if the midi is the same note with the same velocity the samples that are triggered will vary. If you took those samples from a real drum track it would sound very similar to what you posted. That's why I asked if is was a soft synth or not.

Where is this "manual" gating coming from? A plug-in? Did you cut out the space between hits and add fades to the resultant clips? Something else? The initial transient of that 11th hit is only about 134 samples at 88.2. So depending on your buffer size the difference in the way Cubase's rendering engine works on real-time fades and effects at 88.2 vs 44.1 can make a big difference.

I listened and it sounds like some hat start to bleed through the last hits which will add to the spike in volume - Do a Test like this - take one single hit from that track and loop it -see how that works .

1. Whats on the main buss?
2. Are you subgrouping the output of the snare drum then its going to the main out?
3. Are you dithering on the main buss?

Trying to think what could be giving you this issue

especially since it could be using a different sample at different times (multi samples)

To fair to the OP and I agree about slightly "full on" tone of his posts however, since updating Cubase last week i have noticed a couple of strange things occurring with levels, particularly with drum transients. The peak levels are tending to vary from playback to playback by a couple of db here and there. Given I am doing zilch, nada, to alter levels or the mix between playbacks it has had me a tad confused. What's more it seems to be something that is intermittent, ie won't happen for a couple of hours and then decides to do it. The effect, to my ears, sounds like a double triggering of the same hit, the bit that truly has me wondering, is that, it's always the same hits that alter. The one thing that does come to mind is that, as I'm using a multi sample snare replacement that, Cubase is sending at times, a random false "double command" for s ingle midi note and the drum programme sees it as "two hits" and uses two samples thus, giving the extra level.

they all sound different to me,
a perceived difference, I hear it plain as day between most of them,
I've done my tests, my ears can pick it blind, multiple times....I PROMISE YOU I can locate them blind, in the room,
logic, reaper, dp, pro tools, pro tools HD TDM, Pro tools HDX, cubase/noobendo, harrison mix buss [....on purpose], I could continue.

Unpopular thing to say here. Get your flame coat on.....