Ableton Live VST Instruments Accuracy Issues?

[Mineral]

I'm experiencing timing differences with audio and MIDI tracks, I've posted a video demonstrating the problem, is it a known bug?

Link: https://www.youtube.com/watch?v=Rp_cosYDZyA

[Angstrom]

Hmm, I watched the video and thought it seemed odd so I tried it out and ... mine is totally fine.

I'm on L10.1b14 , windows system.

What I did.

1. Made a track with an Operator Kick drum noise playing 4 beats to the bar, for 8 bars
2. Duplicated that, froze and flattened the duplicated version
3. lowered both tracks to -3db
4. put a Utility invert on the flattened track
5. played it back

Outcome: total silence with no rogue transients.

So, I've no idea what's happening with your setup there!

[Angstrom]

I did a quick screen capture to show that it seems normal on mine
https://www.youtube.com/watch?v=CjZZJLX1EPw

you'll hear one rogue "thud" as I turn off the phase inverter for a beat to show that it is actually making sounds, but that they are all being phase aligned.

[fx23]

@angstom, try do the same test at say 121 Bpm, and try to loop the audio clip/arrangement, here 120 is fine, but 121, 22,23 ect are not

my pers findings is that tempis that fit integer number of ms per beat will fully cancel (ie 60,120,150..) but ANY other will drift and won't have
sample accuracy.
that seems to point there is a precision/accuracy flaw with audio clips computations rounded to ms precision, or internal system clock query lack of precision.

[Angstrom]

here's a test at 121 BPM. there is an audible discrepancy on the first beat but as far as I can tell everything else is fine.

I'm not saying "there's no problem", only perhaps that it is more complex than it first appears.

https://www.youtube.com/watch?v=7n4sDVoWTe0

[fx23]

this is especially more visible in resampling with short audio clip looping, ie here one bar:
first is 151 bpm, it clearly doesn't cancel and the drift is random each clip looping turn,
second is 150, perfect cancel, no drift, -inf on master. this reproduce to all ms fitting bpm

https://youtu.be/CxWgEP-66O8

[Mineral]

this is especially more visible in resampling with short audio clip looping, ie here one bar:
first is 151 bpm, it clearly doesn't cancel and the drift is random each clip looping turn,
second is 150, perfect cancel, no drift, -inf on master. this reproduce to all ms fitting bpm

https://youtu.be/CxWgEP-66O8

Wonder if this is a minor bug or a fundamental issue

[fx23]

**correction its integer nb of samples fitting tempis rather than ms.

in a way mathematically speaking,most tempis will make loops not fitting an integer nb of samples, and so kinda 'normal' to be rounded/truncated/different durations loops each turn, however all should be affected in the same way to my mind if trigged at same requested start time
and of same defined beat durations.

even if say a short midi 16th clip is copy pasted on the timeline each beat, and its wav bounced version is too, even if they are not warped/looped, they won't trigger and remain consistent wich imo is def not a normal behaviour.