Aeros 3.4.2 - Tracks of the same part shifting over time

Problem statement

If you let a track loop long enough, unrecorded tracks of the same part shift relatively with time. The more you wait before recording empty track the more it shifts.

Context

  • Aeros connected to BB (so in quantized mode)
  • Aeros firmware 3.4.2
  • 2x2

steps to reproduce

  1. I trigger the recording from the BB
  2. I record the first track, everything’s aligned
  3. I jam along… :guitar: :timer_clock: second track (unused) starts shifting
  4. After few minutes the two tracks have really shifted
  5. If after that I try to record something… Welcome to the twillight zone…
  6. Result quite… interesting. Notice the red bar is now where the track was shifted, and the white bars shifted the opposite side… :stuck_out_tongue_winking_eye:

Conclusion

This is the first time I find a consistent way to reproduce this one. I believe it’s there before the 3.4.2 release but not completely sure… Nevertheless this topic supersedes my point 3 in my remark in this 3.4.2 release sheet.

I didn’t try without the BB in the but I doubt it plays a role.

1 Like

I have noticed this before too! Where I’d record first track, then 2nd track the red line would be off just a bit like in your last pic! However I can’t recall if it did this before in previous firmware or just this new one. Also I don’t have or use beat buddy, this is happening to me just on aeros by itself but I do have a morningstar MC8 hooked up too it that I just got a few weeks ago. Maybe it has something to do with midi?

Was doing some frippertronic stuff tonight and got the same issue again with things not aligning right… however if i stop the recording and then resume they usually snap right back correctly in place.

1 Like

Oh brother WTH is going on here. I won’t make it back to the studio for a couple of days to help test, but this is concerning. And I’m eager to confirm.

I find it interesting that Laurent’s track 2 is slightly ahead where Brock’s tracks are falling behind. Thats bizarre.

Edit: Is this simply a graphical anomaly or are the tracks audibly drifting as well?

I assume sync start and sync length are both active?

I don’t think so, everything related to MIDI sync is at the song level.

That’s the point ! When you are syncing with a MIDI device, those parameters are set by default (and grayed so that you cannot change them, at least when using BB). I believe the problem is at this level, I mean the problem seems related to the syncing algorithms driven by these parameters.

I think I can give (more or less) good news on that front:

  1. I record something and wait until I can see track 2 clearly shifting (few minutes):
  2. I record something as second track easy to match with the first track waveform:
    Again notice that after having recorded the second track, it looks like the the shift side of the second track reverted (or mirrored with zero being the first track). Hopefully it is clear on the pictures.
  3. The good news, After a long wait letting the loop run, I see no additional shift, therefore the two audio tracks remain in sync :+1:.

Although, it may then look like a display glitch only, it is very disturbing, and the audio seam is not very clean, it looks like the beginning and the end somehow overlap… maybe due to this inversion of the shift side after a record of the second track.

@BrennanSingularSound I think it deserves the name of serious bug. And the more I analyze it the more I think it has been introduced with the latest firmware, or at least its severity increased, I seriously doubt I would have missed it. Maybe someone still on the 3.3.x could help there…

I can add to this that it only occurs if you record one track and do not immediately record the next one (maybe @Brockstar could confirm that in 6x6 the behavior is the same), as it seems recording a tracks freezes its shift with the previous track and shifting takes a bit of time.

This is why it’s not that easy to spot in a “normal” usage. I will edit the original post to reflect this remark.

Correct, if i record fast parts from 1 to 6 the shifting doesn’t seem to happen, only seems to happen when you record, let it play for a bit then record another eventually and let it play and then record another etc… However in 6x6 mode I really don’t see the shifting happen until around the 4th track for some reason. Track 1, 2 and 3 seems to be stable but once getting to 4th track that is when it’s starts shifting for track 4, 5 and 6. I just did few more tests this morning and it did the same thing each time. And again NOTE: If I stop the entire playback and resume, it all snaps back into place.

2 Likes

This is true for me as well, until I try to overdub or re-record… then I see again the the shift (although it does not affect audio).

I’ve had that happen before. Never could reproduce it. It happened randomly. Very frustrating.

2 Likes

I am coming across some serious bugs in 3.4.2 that may have been there before, I will have to do some more testing then go back to 3.3.3 etc.
On recording say one part 4 tracks. Track 1 and 2 are 8 bars Track 3 is of a much longer length say 30 which is an uneven multiple. Then recording track 4 which is again 11 bars of uneven multiple. There happened to be a gap in track 4 with no sound, I then did an overdub in track 4 to fill in the gap. The overdub changes to green, but on playback there is no sound from the green section.
I will have to try and reproduce this in a more simplistic way. Then go back to an older firmware.
The bars showing uneven alignment do not seem to effect the timing of the tracks.

@Johnyhorse Do not hesitate to create a dedicated topic and link it to the 3.4.2 release sheet in order to be able track issues.

Same here. I’ve had this happen with the prior version too, but I don’t recall exactly what leads to it happening. I use external MIDI Clock and I kind of think in the prior version it had something to do with tempos not being set exactly the same in both the sending device and Aeros (could have been a combination of that with certain sync settings), but everything is okay if I set them both and normally just stick to what I know works. Sounds like the bug is expanding.

I can look into reproducing it in the prior version if anyone thinks it can help, I’m avoiding doing that right now because it may just add confusion.

A little more self testing results… Ok so I guess by 3rd track it starts to shift. I thought it was 4th, 5th and 6th, but it begins slightly on 3rd track. Also I noticed that if I click “next track” and it pops up and getting ready to record when the measure of the last track hits but if i bypass it and go back to track 1, then 2, then 3, then 4 sometimes the alignment will snap back in place 2nd time around on recording next track or sometimes it’s only just a hair off but not as bad as before.

Ok this is all very useful! We have long seen versions of shifting happening, it is great to find a reproducible action to getting this behavior, we will definitely be looking into this, thank you. It must have something to do with 1st track versus 2nd+ track, and 2x2 versus 6x6, we’ll have to look into scrolling logic.

To be clear, If you do not have locked tracks and the part is not switched right “on the beat”(within our loop forgiveness zone, about 500 ms where if you miss the 1 you still switch to the next part, not wait for the next sync point), are you hearing the desynchronization?

To be clearer, if you switch parts well before the measure line, do you hear a desync? Or is it purely aesthetic?

Thanks

1 Like

This is slightly inaccurate, you can change the setting as long as it has not been recorded to, are you seeing the setting unchangeable even when you clear the song? Thanks!

Ah ok thanks for the precision, I didn’t pay attention to this point. It’s out of scope but good to know,

No part part switching involved at all, no locked track, no “on the beat” issue.
Everything happens within the same part with no part change, with the exact description I’ve made.

Same for me.