AudioSnap Detection Bug

[Jakub Maggi Malášek]

Hello,

i found bug with detection of starting point of the curve, in this case is Kick Hit... - Almost All of Kick hits are off spot...

Same waw track opened in old 8.5 Producer and all hit are on spot. This needs to be fixed.

You cant edit drums  without bleeding your hands on the mouse to do manualy every hit... 

[Lord Tim]

Audiosnap is a great tool that works really well ... once you get past the detection issues. The big problem with the detection at the moment is it's bad for false transients if there's any noise around the hit. Sometimes you get markers early, sometimes they end up in the middle of the hit. Both need a lot of manual work before you can actually work with them properly.

Back on the old forum, I posted up a few steps that got past that issue: http://forum.cakewalk.com/Tightening-multitracked-real-drumkit-by-using-AudioSnap-anyone-using-on-regular-basis-m3387974.aspx#3388204  (scroll down to reply #7)

It's a bit of mucking about but the results will save you hours of manual work, and CbB uses a FAR better stretching algorithm (if you use stretching rather than splits) than SONAR ever did, so the end result sounds excellent.

[Byron Dickens]

Or, you could get a drummer who can play in time....

[Jakub Maggi Malášek]

Lord Tim - but in new CBB is every hit off right spot... Same project opened in Sonar 8.5 and all hits are on right spot so there is problem with that  My tracks are pretty clean  I use this for triggering and for that is crucial... 

Edited June 27, 2020 by Jakub Maggi Malášek

[Jakub Maggi Malášek]

27 minutes ago, bdickens said:

Or, you could get a drummer who can play in time....

Yea but also customer is customer... I need to pay my rent also  

[Jakub Maggi Malášek]

22 hours ago, Lord Tim said:

Audiosnap is a great tool that works really well ... once you get past the detection issues. The big problem with the detection at the moment is it's bad for false transients if there's any noise around the hit. Sometimes you get markers early, sometimes they end up in the middle of the hit. Both need a lot of manual work before you can actually work with them properly.

Back on the old forum, I posted up a few steps that got past that issue: http://forum.cakewalk.com/Tightening-multitracked-real-drumkit-by-using-AudioSnap-anyone-using-on-regular-basis-m3387974.aspx#3388204  (scroll down to reply #7)

It's a bit of mucking about but the results will save you hours of manual work, and CbB uses a FAR better stretching algorithm (if you use stretching rather than splits) than SONAR ever did, so the end result sounds excellent.

Lord Tim - but in new CBB is every hit off right spot... Same project opened in Sonar 8.5 and all hits are on right spot so there is problem with that  My tracks are pretty clean  I use this for triggering and for that is crucial... I will check your style of editing. But thats not what i need to fix but thx  

[Lord Tim]

I can see in your screenshots that there's still some minor noise around the hits. Gates or not, they're not clean as a drum machine sample, and even then, you'll sometimes get false detections during decays, so aggressively gating, bouncing to a new clips so the actual source itself is clean and then using that as your detection track is the best way to save you unnecessary work. It's absolutely that picky.

As I said in my older post, the detection algorithm was changed since 8.5 and the side effect was definitely later hits on drum detections, although it did improve detection for other material quite a bit.

I completely agree, there's no question that you shouldn't need to adjust your workflow to get around this issue - it should be fixed (and not just reverting back to the old algorithm, the reason that was changed was it fell over on some material too). I know at least one person (ie: me) has reported this issue and the devs are aware of it, so it wouldn't surprise me if adding pre-filtering is already on the to-do list once they get a chance.

But I've always been one of those guys where if something isn't working, sure - get annoyed, definitely report it - but then see if you can find a workaround in the meantime so you can get the job done, which is what my old post was about.

 

[Jakub Maggi Malášek]

On 6/27/2020 at 7:22 PM, Lord Tim said:

I can see in your screenshots that there's still some minor noise around the hits. Gates or not, they're not clean as a drum machine sample, and even then, you'll sometimes get false detections during decays, so aggressively gating, bouncing to a new clips so the actual source itself is clean and then using that as your detection track is the best way to save you unnecessary work. It's absolutely that picky.

As I said in my older post, the detection algorithm was changed since 8.5 and the side effect was definitely later hits on drum detections, although it did improve detection for other material quite a bit.

I completely agree, there's no question that you shouldn't need to adjust your workflow to get around this issue - it should be fixed (and not just reverting back to the old algorithm, the reason that was changed was it fell over on some material too). I know at least one person (ie: me) has reported this issue and the devs are aware of it, so it wouldn't surprise me if adding pre-filtering is already on the to-do list once they get a chance.

But I've always been one of those guys where if something isn't working, sure - get annoyed, definitely report it - but then see if you can find a workaround in the meantime so you can get the job done, which is what my old post was about.

 

Is there some offical way to report this ,,bug,,? Cant find any contact directly to CbB  

[Lord Tim]

Yep, shoot them a line at support@cakewalk.com