[SOLVED] Bug (incl. sample): Audio Warp point quantizing

SOLUTION:

Thanks for the dude which name I forgot but he has written to either this thread or the other (linked below).

One has to first calculate hit points like I did, then change them to warp points and doo quantizing like I did, BUT what has to be done is to change the algorithm for drums to STANDARD -> DRUMS.

Yes, pretty logical now that I think of it but another question emerges; Why the heck does it not work with Elastique algorithm? Shouldn’t it? Well, STANDARD - DRUMS algorithm seems pretty good quality to my ears as I listened to it - and - because it is an algorithm specifically designed for drums, then I accept this solution. At least for this particular project since it solved my problem.

However I wonder what does that Elastique algo do to the audio then? When should it be used and when not? In what situations what algorith is the best? Since you might wanna quantize bass guitar and vocals etc. manually so should the DRUMS algorithm be used for those too?? Seems very strange to me…

What I would LIKE to use is Elastique in EVERY situation. There are formant times, formant pitch etc there so I don’t get that when should these STANDARD algorithms be used in the first place. Well, in this particular situation apparently this is exactly how we must do :slight_smile:

Anyway, I’m glad that the problem is solved. Still fuzzy about the different algorithms thou…


ORIGINAL POST:

OK, I have this one project of my client and a song named Relief, that has been played (drums) so badly that I have to make a hitpoint quantizing to the drums.

So what I did is as follows:

  • I double clicked the already downmixed drum track and selected AUDIO - HITPOINTS - CALCULATE HITPOINTS
  • Then I selected: AUDIO - REALTIME PROCESSING - MAKE WARP TABS FROM HITPOINTS
  • Then I quantized them with Quantizing Panel.

OK, everything seems fine graphically when viewed on the multitrack. All the drums’ hitpoints seem to be in place. But Cubase won’t play NOR downmix the drums correctly!

Here is what I got: http://www.pc-professor.fi/Cubase6/test.mp3.

It starts nice - well as nice as it can be considering how the drums were recorded - but when the song goes along, the drum beat offset is starting to emerge. And finally it’s completely offcue !! As you can clearly hear. It’s actually much WORSE than the original recording!

And yes, there are individual hitpoints wrong in that drum file, don’t pay attention on those. I yet haven’t manually fixed it. But the overall - or offset positioning of the hitpoints is totally screwed up! Not when viewed by my own eyes from multitrack or sample editor but only when it is played from start to finish.

The weird thing is that if I hit STOP and then PLAY again from that stop point, the drums are perfect again - but only for awhile. They start going offset again slowly, being completely wrong in the end.

I have tried musical modes, linear modes, tempo is correctly adjusted both from the pool for individual tracks and it matches the project tempo.

Yes and one thing; the drums HAVE BEEN originally recorded with click (130bpm) and they are somewhat right originally but this hitpoint quantizing messes them up even more!

It was the same thing with Cubase 6.05 and I hoped the Cubase 6.5 would finally fix the issue, but it doesn’t. So am I doing something wrong here?? PLEASE HELP!!!

And yes, I have tried “flattening” the audio, bouncing, downmixing only the drum track with warp quantizing enabled - nothing works. It totally screwed up.

C’mon please help somebody! Helge??

Hello, that’s exactly what happens to me sometimes! But only sometimes - I wasn’t able to isolate the problem yet but I have some suspicion. What stretching algorithm have you set up for that track? Do you use any signature changes in this project?

Anyway, I suppose there is just one workaround in this terrible situation: downmix the warped track stepwise by small chunks spanning only several bars. Then crossfade the resulting chunks together again. Nothing else worked for me… :frowning:

Best regards

Miloslav

Thank you for the post! OK ok, it seems like I’m doing everything right then if someone else is having a same problem. More or less, I think this BUG is a pathetic display of Steinberg’s engineer’s coding abilities. I’m definitely not going to start downmixing my project piece by piece. i actually thought of that too, but it soooo unelegant that if Steinberg cannot come up with a solution then I only have to wonder what I have paid for!

Come on, Steinberg people!! Others have the very same problem and shouldn’t you actually do something to make your software work how it’s designed? What use are these fancy features for if they don’t even work?! Makes me wonder how many other bugs this software has…

I mean the purpose of Steinberg’s attempts to make a perfect DAW are good, but the implementation totally sucks ass! I hope you at Steinberg don’t start saying something like “Well, it was not designed to work in that extent”. Damn it that I’m pissed. how the hell haven’t you noticed before by yourselves???

Thank you, Steinberg!

The new hitpoint quantizing thing is a great thing - it just doesn’t work!!! Thank you! You made a good effort so thank you thank you! It’s no use of making it work well, since it doesn’t work at all now. It’s really not - I only feel great to have such an expensive piece of software with so many features that don’t work! It makes me feel good!

So thank you for the new “mess up quantizing to be even worse than without it”-feature! Now I REALLY can make this awful sounding mixings that I always wanted to!

All the other DAW’s probably would work better but their outcome is annoyingly flawless that the song won’t stick out of the crowd. Well, now that I can use Cubase to ass-quantize my audio, I REALLY can give my clients a possibility to stand out!

Thank you!!!

I feel your frustration but emotions do not have impact on Steinberg! :wink:

Honestly, I think that Cubase is already very complex and the more complex it become the more difficult it is to debug it. Maybe this bug occurs only in very special conditions. Otherwise, people would report it more frequently, I suppose.

Anyway, it would be great if we could just send the affected project to Steinberg’s testers and developers. So, please, moderators, is there such a possibility?


Miloslav

So Tommy - can you supply the ‘before’ downmix drum track to SB to look at themselves…? Have you raised a support ticket…? How long is the actual song (just curious, its not like 18 minutes or anything…! Ha!).

Something definitely seems weird about your described scenario…

You say everything ‘looked’ right - but then you say some of the hitpoint detection wasn’t accurate, which you know you can fix; so its the general malaise of the ‘drift’ of the groove that shouldn’t be happening - especially if later on all the hitpoints/warp tabs LOOK good after quantising, against the grid… I mean, thats just not right…

I may get a chance over the weekend to give this a go myself, see what occurs… (as they say in Wales…!)

bob

I think your chances of getting a resolution will increase by 100% if you can share some version of the project out to Steinberg or (more probably) the forum here, even if you have to cut most of the stuff out of it first.

Additionally, it will serve as proof that you’re not batsh!t crazy.

Hey there!! do you have any linked channel in your mixer? If you do, unlink the channels and quantize again. Let me know!

Hi!
Can you send us the project? It’s entirely possible that a potential issue have been overlooked… that happens since software is written and the Bugs are around. Sure, I know this doesn’t help you right now in the situation, just for the record.

If we can reproduce this issue, we aim to fix it in the next maintenance update.

Best regards,
Helge

Do we say that in Wales ? Not sure i’ve used it :laughing:

Hi, Helge! Thank GOD !! :slight_smile:

I will most definitely send you the complete project with all the audio files etc. etc. Well not everything has to be sent of course but you know what? I’ll put the files on my web server and share the links. Actually I can quickly make a small web page with all the pictures and linked files etc. Whatever has to be done to make Cubase even better! I will do it.

Regardless of my earlier frustrated mails I do want to help since I love Cubase. Too bad that some features don’t work properly. Cannot name anything else thou, since this is the only issue that doesn’t work.

AND I want to emphasize one thing::: When the playback is stopped AND the playback is continued from that very same spot where STOP was pressed, then the drums are perfectly in place - but only for awhile. It seems like some kind of an offset is accumulating over time. It really seems that way. Honestly I can say that I have tried everything. But isn’t the fact that audio warping is fine after restarting the playback some kind of an indication of a bug? I think it is.

So stay tuned my dear friends. I’ll put up all the files next thing tomorrow morning. Now I’m beat and off to bed…

-Tommy

Seems you’ve got a good point there.

…and I thought you were just crazy. Shows what I know.

Curious additionally, so then if you set your loop start somewhere in the middle of the project and render offline, it will lag some time after the loop start in the exported render like it does if you had started there on playback. (?)

Yes, I will demonstrate all that. I’ll start doing a web page for that right away.

And thanks for not thinking I’m crazy anymore :smiley:

Check out this web page of mine: http://pc-professor.fi/Cubase6/warp/warp-quantizing-bug.html

It will demontrate you all from scratch how the quantizing fails.

Also check this thread: https://www.steinberg.net/forum/viewtopic.php?f=19&t=19982 !!!

I also posted all the info to Helge Vogt so this issue is under investigation now. If someone missed it, Helge himself posted on this topic.

I did a test yesterday with 8 tracks of drums using the multitrack quantizing method that’s described in the manual and it worked without a hitch. I don’t quantize much and am far from an expert on hitpoints so I’m a little fuzzy on the differences between you method and the method I used which was…

Analyze the kik snare and overheads, group all the tracks, click the quantize window and cut, quantize and crossfade the tracks.

Tried your project and could reproduce with “élastique pro ->time” algorithm, however switching to “standard->drums” solved the issue.
Also, it’s not only starting and stopping playback that puts the drums back in correct timing for me, opening the audio editor or toggeling between musical and linear time base while playing works.

/A

Yes, I know it works when changing on some attribute - I just didn’t mention it on the page. It was its own job to compile the actual page so that it would be clear to everyone what I’m talking about :slight_smile:

But - changing the algorithm to STANDARD - DRUMS worked? WOW! But STILL - shouldn’t it work also with Elastique since isn’t Elastique the “main” algorithm to use with? And if it doesn’t work with Elastique, then why not? I would think that audio is audio regardless of the algorithm used.

Anyway it seems to me like some kind of a bug. I mean what is the use of Elastique algorithm if it fails to work properly under all conditions? Namely I’m a bit afraid that the quantized drums using STANDARD / DRUMS algorithm might sound bad but I’m not sure… I’ve got to try it right away.

Helge, what do you think?

Anyway, thanks for this and I’ll return here if I fail to do it properly.

Of course this should work with élastique, and there is obviously a bug specific to that algorithm, especially since the quantizing is visually correct.

But don’t be afraid to use the standard algorithms just because they are named like that, the “standard-drums” sounds miles better than the “élastique-time” that is pre-set on the audio I received from you for instance.

edit I suggest that you remove [SOLVED] from the title and add “… using élastique” for clarification.

/A