Hi dear Gearslutz people,I recently discovered a serious bug affecting Cubase's (I don't know in previous versions but I guess it's not different) v8.05 (and other DAWs) Export Audio and Rendering in place functions.

So I decided to report this issue in the Steinberg forum where it has been previously approved by moderators, and then removed for no understandable reason and without me having been advised in any manner.

The thread has been there for 2 days with more than 50 views and no answer from steinberg or users before being silently removed: underwhelming to say the least.

Here a copy of my newly submitted (pretty much equal to my original issue report in all of its relevant parts) Issue Report including
-a .zip containing a Cubase project folder useful to reproduce the bug
-a picture clearly showing the bug and
-a .zip file containing a .gif that clearly shows my newly submitted request in all of its parts and it clearly shows that the request was correctly sented to Steinberg and it is
waiting to be approved by moderators.
I did this just to clarify that I did not miss anything when submitting my request and I respected all the rules that have to be followed for posting Issue Reports: now they can remove/not publish my thread but I can prove to anyone that they shouldn't have done it.

Hopefully it was only a technical problem with their support/platform and the original thread wil be republished but in the meantime ...here it is:

UPDATE:my thread has been re-approved and the following request disapproved so you won't find it on the Steinberg Forum.Here is the link
for the re-approved thread: http://www.steinberg.net/forums/view...?f=230&t=75106
-------------------------------------------------------------------

"Hi, my topic concerning this very annoying problem has been removed without any reason from the "Issue reports section" of the Steinberg Cubase 8 Forum where it was previously correctly appearing after having been approved by moderators.

This http://www.steinberg.net/forums/view...30&t=75106&e=0 was the address of my previously approved post that now, as you can see for yourself, is no more accessible.

Can I have an explanation for why this has occurred?

I followed all the rules stated here:

www.steinberg.net • View topic - How to report a problem with Cubase

1-I used the search function to be sure that this problem wasn't being already discussed/solved in the Forum and of course

I searched in the manual for information about this without finding anything useful.

2-I only did include one issue in my report

3-The thread title reflected a short description of the issue

4-I posted multiple steps to reproduce the issue and I even included a .rar file containing a project where you could reproduce the bug and a picture showing the added chunk of audio at the end of the audio file.

5-I correctly specified my system info,Cubase version and what 3rd party plugin I was (not) using (no vst loaded in the project).

Also, I used a moderate language so I can't really understand why my thread has been removed without even a PM advising me of that and explaining me why.

Am I missing something?

Anyway I saved as a document this entire post ,so just in case it will be deliberately removed again I can be sure that this time I will not lose my report and I will be able to post it again without losing much time.

Let's start again...

START OF BUG DESCRIPTION:

My system specs: Win 7 Pro 64 bit SP1, Cubase Pro 8.05 , no 3rd party Vst plugin involved.

As the title says I identified and can reproduce a bug occurring in "Render in place", "Export selected audio tracks",
"Export Audio Mixdown" and that is being discussed by me and other Cubase users here:

www.steinberg.net • View topic - Render in place bug:rendered item longer than original item.
www.steinberg.net • View topic - extra audio length on exporting audio on cubase 8
www.steinberg.net • View topic - render in place bounces extra audio

The bug consists in this:

1 Create an empty project and set your project samplerate to 44100.

2 add 1 audio track

3 import an audio file of your choice(longer than 1/16th note at your project tempo)at the start of the timeline ( 1.1.1.0)

4 select the end of audio file and reduce its length to 1/16th (set end to 1.1.2.0)

5 set locators to start (1.1.1.0) and end (1.1.2.0) points

6 make sure that you have tail length set to zero in Render settings

7 Do a "Render in place", "Export selected audio tracks"or "Export Audio Mixdown".

8 Import rendered track in a new audio track at position 1.1.1.0

9 Set the cursor at the end of selection (1.1.2.0)

10 Zoom in at maximum possible zoom where you can see the single samples drawed like rectangles on the timeline (use the

"H" letter on your keyboard to do this).

11 Observe

You should now see that the rendered file length is different from original file:Cubase will add an extra chunk of audio

with the size of 1 sample (circa?).

This behaviour can cause all sort of timing/phasing problems when duplicating/replicating items if you don't pay much

attention to be sure that everything is in the right place.

I think this problem is strictly related to sample rounding errors (samples not alligning to grid),a problem that affects

other Daws including Ableton Live 9 (warp mode off), Reaper 4 and Pro Tools.
More info here: PT and Sample Rounding WILL cause Issues when Copying - Avid Pro Audio Community , Duplicate/Copy&Paste errors with Samples on Grid - Avid Pro Audio Community ,

Rounding Errors (?) [Archive] - Cockos Confederated Forums and several other posts that you can google for yourself.

Using a submultiple of the currently used sample-rate as the project tempo doesn't help to fix the problem.

This is clearly a serious bug/not wanted behaviour and should be addressed as soon as possible and at least clearly
mentioned in the Manual (Pro Tools does this ) and/or in the "Version history and known Issues" section of Cubase in the Steinberg website.

What do you guys think about this?

Any information, opinion or hopefully a solution is much appreciated.

An official reply from Steinberg would be much appreciated and needed IMO.

Thanks for your attention."

I'm very underwhelmed from this situation and I'm beginning to lose my trust in Steinberg after this bad experience: it seems that the voices regarding Steinberg's horrible forum support experiences reported by many users here and on other platforms are more than just plain rumors, I hope that at least here this post will remain visible for a long time


Please share your infos/thoughts/solutions about this 2 problem guys: the software one and the support one

Bump.

Don´t understand point 4 in your explanation.
What is the sense of cutting the original file to any measure and then rendering it ?

What BPM is your project at?
From my understanding this is a rounding error between the BPM and the audio clock rather than an actual bug.

You might argue that the internal timing resolution should be higher but as we can not have infinite resolution, there will always be a rounding error. The question is what forms a good compromise between timing accuracy and DAW efficiency...

There are other ways to demonstrate such rounding errors and as far as I know they occur in every DAW.

Alistair

My BPM was set to 120 at 44100: I choosed it with the intention to actually discover if there was any sample rounding error.

BUT the same problem (in Cubase) occurs even at 126 or 147 BPM that are exact submultiples of 44100.

In Logic if using exact submultiples of 44100 like:
87.5, 88.2, 90, 94.5, 98, 100, 105, 110.25, 117.6, 122.5, 126, 131.25 etc..
the sample rounding error will not occurr.

In Ableton Live is more complicated,it doesn't follow a rigid pattern i.e.:
126 = no sample rounding errors
110.25 = no sample rounding errors.
88.2 = sample rounding error occurs.

But the more important is that from my tests Sonar Artist (last version) doesn't exhibit this behaviour!.

I tried several times to reproduce this "bug" in Sonar but it seems to work ok.
It should be noted that Sonar dislays the Grid in MIDI ticks (240,480,720 ecc.. instead of 1.1.1.0 , 1.1.2.0, 1.1.3.0 ecc...).
Maybe this has some relation to the appearently correct behaviour of Sonar?

Is there any Sonar power-user who can confirm me that Sonar doesn't suffer from sample-rounding errors? (I bought a 1 month sonar artist subscription just to test for this problem so I'm not really so confident with that DAW to say that everything is really working fine there).

I'll also add that in Logic even if there's the sample rounding error in the rendered file( when working with non submultiples BPMs), when duplicating/repeating everything works fine.

These are the reasons why I'm not sure if I'm missing something in Cubase.
If 1 DAW (Sonar) seem to work ok at least from my tests and I couldn't find any issue report about this subject on Cakewalk board and here on GS, maybe there's some solution that I'm not aware of to avoid the problem In Cubase too?

And I'm also curious to understand why (if my tests are correct):

-in Logic whe duplicating/repeating everything works fine at any tempo even when there's a sample rounding error
-in Logic when using exact submultiples the rounding error will never occurr in contrast of Ableton Live in which seems to occurr random and in contrast of Cubase in which appears to occurr always even when using exact submultiples.

In this case I did it to test for sample rounding errors, but the main reason is for looping/duplicating/repeating tasks where I want my item length perfectly alligned With the Grid to avoid timing/phasing inaccuracies that can occur when duplicating the item over the timeline.

Interesting SilverSheep. I'll do my own tests when I have time.

In previous tests I have done I found that I got timing rounding errors when rendering soft synths except at 120 BPM. That is why I asked.

Anyway, one thing you could test is to set the loop by using the "Set locators to selection" function instead of by hand. (Select the clip and hit p if you still have the default key commands set). And then render like that.

Alistair

Thanks Undertow, tomorrow I'll try your suggestion and I'll report my findings here.

I'm assuming that when you got no sample rounding errors at 120 BPMs you were working at 48000 hz Samplerate and not 44100. 120 is not an exact submultiple of 44100 but of course it is a submultiple
of 48k or 96k, hence my assumption.
I didn't try tests at 48000 Hz Samplerate if not in Logic where everything worked as expected at 120 BPM/48000 Hz.

Just for additional information I'll add that I also did extensive testings on this subject in Reaper, a DAW with which I'm very confident, and in several days of meticulously done tests (which of course could still contain some kind of user error which I'm not aware of) I tried every single possible combination of settings which affect timing in that DAW:
"Snap to sample rate", "Snap relative to grid", "Force tempo/time signature changes to occur on whole samples" and using submultiples of samplerate
but NONE of these settings' combinations gave me reliable results.
When I say "every single possible combination of settings" I mean ALL of the possible combinations, like: ON,ON,ON / OFF,OFF,OFF / ON,OFF,OFF /ON,ON,OFF etc... and I'll repeat it: NONE of these settings work with reliable results.In example:
-Even when the length of the rendered audio file is correctly represented , as is the case with the 3 settings that I described above ("Snap to sample rate", "Snap relative to grid", "Force tempo/time signature changes to occur on whole samples") all ON,( in Reaper they call it "sample accurate mode")
when duplicating/repeating the rendered item, there will still be sample rounding errors, even when using submultiples of samplerate as 126 BPM at 44100. (at least in my tests, that however, seem confirmed by other Reaper users as well).

I even went further and I tried to do a null test in Reaper to see what was happening there between original and rendered file in Reaper's "sample accurate mode".
I wanted to see if the original and rendered file, which appear to have the same length but don't duplicate /repeat correctly, (as is the case in "sample accurate mode" see above for more info on this mode) nulled (at least till before the last sample): this is NOT the case.

Strangely the 2 files will null from -144 to -138 db ( depending from position of the playback on the timeline ) and not to -inf, even when starting the play from a random position on the timeline ( to exclude possible starting/ending rounding errors which could affect the meter reading ).

It should be noted that I had no crossfades/autocrossfades activated and my bit depth was set to 32 bit for recording Audio.
I was using Reaper v4.75 x64 on a Win 7 Pro SP1 system.

It should be noted that in the aforementioned Reaper's "sample accurate mode" even if the length of the rendered file appears correct, when you zoom in at the sample level, you can clearly see that at least 1 sample is not written at all and there's at least 1 sample of non written/silence data at the end of the recorded item. This is something I sincerely would not want and I want to be informed if this is a known behaviour/issue.

I'll do further testings when I have more time and report here for anyone interested.
I can upload gifs of all the tests that I have done , for anyone who hasn't fully understood the problem or simply wants a visual reference when trying to understand the issue.

I think that if we discover/can be sure of at least 1 DAW that does it correctly ( Sonar? )or that however work around the problem in a reliable and quick way when duplicating/repeating ( Logic? ) , then we could have a proof that not all DAWs have this behaviour and move accordingly.

Having this in mind when talking with developers and even users about this issue will hopefully force all (professional) DAW developers to create a universal and reliable solution to this problem and in the meantime they could inform all customers in a clear way that there are this kind of issues when dealing with current versions of their DAWs.

AFAIK only ProTools does at least mention of this issue in the manual and offers a solution :Tick based tracks, that, however, still don't work, as you can read here :
PT and Sample Rounding WILL cause Issues when Copying - Avid Pro Audio Community
and here:
DigiZine and from other sources if you do a Google search.

(I didn't try it for myself in Pro Tools because I don't own that DAW so I'm assuming that the infos contained in the links above are correct but I can't be empirically sure of that).

I'l wait for your reports UnderTow and I'll suggest to anyone, especially Sonar users to do these kind of tests and report/discuss your findings/thoughts about this issue here.

Thanks for your attention!

It indeed extends an audio file,when you resize it.
The Render nulls to infinity, but you are in trouble when you think that you just did resize a file for looping.

Hi lllubi, are you talking about Cubase here? I'm asking because I did null tests only in Reaper and not in Cubase.

Interesting.

I have sent a ticket with big problems about chord track and midi loops.
You can audition a loop with linking it trough the chord track, but when you import it then it play back completely different!

It is like the worst fail ever..

Hi The Vulture, it's just a few days that I switched to Cubase so I still don't have explored every feature extensively.
I just tried the "chord tracks" and "live transform" features and so far it seems to me a wonderful feature if it can be used in a predictable manner.

Can you surely exclude user error in this case? ( I just don't want to believe that this feature is broken )

Could you give us a step by step guide to reproduce this issue?

This would be useful for people like me that still have not that degree of confidence with the DAW, Cubase in this case, to be sure that what is said is being completely understood, and would save us some time when setting up the DAW to try to reproduce the issue.
Thanks for your report!

Yes, speaking about Cubase 8.
Its a perfect null with with an extended region only if you resize your audio file.

Ok, The fix for multiple channel VIs was to render in place using the "full chain" but I still get multiple tracks in battery rendering (most with no audio). I don't remember this happening before but it is now.