Artifactual or non-generated sound from a Layer near a Part Trigger

VST Live
1

Hi team,

I have been trying to understand for months now to understand why some of my projects seem to occasionally generate poorly or not at all the sound of a Layer with MIDI notes coming from an external instrument coinciding with the start of a Part (position defined by the Part Trigger).

After various hypotheses I have come to the conclusion, which I believe this time is rather evident, that there is a problem with the use of a Shared Layer.

Contrary to what I initially understood, the definitions of BPM, Tempo, and Trigger Part value seem to influence the issue, but not in a completely decisive way; even the fact that the note played near the Trigger Part comes from an external instrument has not proven to be decisive, since even with a MIDI track it is possible to reproduce the issue.

The issue seems to be particularly influenced by the Buffer Size setting of the ASIO Audio Driver, while it seems independent of the VST used in the Layer (at least with the ones I used)..

To demonstrate what I just wrote, I prepared a simple project that involves the execution of a MIDI track that is addressed, before the start play, to an output virtual port Not Shared Layer Port or to a Shared Layer Port: what happens is that the MIDI track is executed either by a Layer instantiated each time in a different Part of the Song, or it is executed by a single Shared Layer.

ArtifactualOrLostMidiNotesOnPartTrigger.zip (729.6 KB)

In the case of execution performed through the Not Shared Layer (obviously the same is obtained using a Global Layer), an execution without any artifact is obtained. as demonstrated by this video with audio:

On the contrary, with the execution of the same MIDI track, but through the Shared Layer, you can notice systematic artifacts that systematically coincide with the position defined by the Part Trigger:

Even if in this video you can hear an artifact sound, but not the silence that replaces the note as happened to me many times with my projects in the rehearsal room, I am quite certain that the issue is the same and that it provides (I hope I am not wrong) a single solution.

I really hope this time I have provided all the information necessary for you to replicate and definitively fix this tormented issue. :roll_eyes:

I look forward to your feedback. :blush:

2

Belive or not,
In a sandbox project, Experienced on monday little pop at part changes with a single audio track.
No fx, no layer of stack. Just a single wav, single little pop each time when reached next part .

Close, reopen, canā€™t repro, sorry :((

3

Hi @fkalmus,

first of all thanks for reporting your experience. :slight_smile:

Out of curiosity, so the project only included the execution of the audio track or were there also some Layer defined in some way? :thinking:

4

Trying to reproduce thisā€¦to no avail so far, but will keep trying, looks like there is something wrong. Thanks for sharing the project!

5

Hi @musicullum,

Thanks for updating me. :blush:

Iā€™ll highlight something that was already said at the beginning of the post, that is, that the issue seems to get worse with the increase of the ASIO Driver Buffer Size.

Also consider that I tried the example project on multiple cards (different from each other) and on more than one PC (only Windows, though) obtaining essentially the same results.

6

Thanks, good hint, will try that.

1 Like
7

Pls check your Part trigger times.
Some are slightly off, but you can only see it when you swith to seconds display.
Set those anew (locate on snapped position and click trigger), better?
Also try with the new version (coming today if all goes well).

image
image382Ɨ405 21.6 KB

8

Hi @musicullum,

I installed version 2.2.12 both at home and in the rehearsal room and ā€œregeneratedā€ the triggers (following your last suggestion) both on my test projects and on my more complex projects.

The feeling is that the issue appears less evident, but that it continues to be present, both in the form of poorly generated note sounds and occasionally in the form of notes not generated at all (this is both with MIDI tracks and with MIDI coming from an external instrument).

I am sure, as already stated, that the issue is linked to the use of the Shared Layer and that it seems to be independent of the VST used for the Layer (at least with the VSTs I tested, namely Battery, Kontakt, Halion, FM8, belonging to the Arturia V Collection X suite), as well as it seems to be independent of the PC and audio cards I use.

I still havenā€™t understood if, even using only the test project provided in this topic, you are able to replicate the issue or not: I repeat that the issue, although always present when reaching the Trigger associated with the Part, becomes more and more evident as the ASIO Buffer Size increases (you could bring the value in samples of your ASIO Buffer Size to the maximum to have maximum evidence).

Considering the fact that I donā€™t see other topics that report exactly what I reported, could it be a problem related to the user data that have been collected on my machines in these 2 years of VST LIve? Consider that, even with new projects just created and without any particular Mixer configuration, the issue occurs.

Doing without the Shared Layer, as the projects I created have been designed over time, would be a hard blow, and, as I had the opportunity to write to you, this issue is holding me back in the path of using the software with the bands I collaborate with.

I await some suggestion or news that can lead to the solution of this frustrating problem. :face_holding_back_tears:

As always, thanks for the support. :blush:

9

Please, first try with the next version. We made some changes in that area.
Should it persist nevertheless feel free to report again.

10

Ok @musicullum.

Iā€™m still curious to know if you managed to reproduce the issue Iā€™m describing, but Iā€™ll also test the next version, really hoping it contains the solution to the problem.

Thanks. :crossed_fingers:

11

Hey @Giuseppe_Loffredo ,

Alghough that isnā€™t a typical setup with VM rewiring, made tests and seems we get a drunk drummer as Buffersize has been increased from 64 to 2048:
https://www.dropbox.com/s/yo6ztjuxi3o4558/vstlive_250304_171250.wav?dl=0

12

Hi @fkalmus,

from what I hear in the audio content you provided, my drummer is teetotal, but we still have in common the artifactual sound coinciding with the definition of the Trigger Part. :wink:

This shows not only that the issue does not concern a particular software/hardware configuration of mine, but that the thing concerns, from what you told me, also the Mac version of VST Live 2.

Thank you infinitely for the help provided! :blush:

1 Like
13

Hi @musicullum,

I also tested version 2.2.13, but the results are similar to those obtained with the versions released in recent months. :face_holding_back_tears:

Since my last communication in this topic, I have asked other users of this forum (also in the Mac environment) to download and run my test project:

https://forums.steinberg.net/uploads/short-url/ttFrzL7YTRfFlmUSVAhdTMAgWNc.zip

The outcome for them too seems to be the same, i.e. sound generation, by a VST inserted in Shared Layer of contiguous Parts, missing or poor near the transition from one Part to the next Part, with the issue getting worse as the ASIO Buffer Size value increases.

@musicullum, I would again ask you the courtesy to tell me if the issue is reproducible by your means, or do you have any difficulty in replicating it: this would help me understand the state of things. :thinking:

Thank you for your efforts.

14

will check again!

1 Like
15

Hi @musicullum,

I continue to receive no response regarding your ability to replicate this long-standing issue, but, despite this, I remain hopeful of a real solution.

As always, thank you for your support.

16

@Giuseppe_Loffredo Could you check one single thing?

Just as last ideaā€¦ Are the WAV file and the Audio Settings on the same sample rate?

Discussed somewhere before, there might be micro crackles when VL needs to convert sample rate on the fly.
(E.g.: tested today with a 44.1 WAV sine file while my audio sample rate was set to 48k, there are micro crackles then)

17

Hi @fkalmus,

premise that I always take great care to match the Sample Rate of the Audio Driver with that of the audio tracks, to avoid, as you rightly observed, useless conversions on the fly. :wink:

Having said this, consider that the issue I have detected, as the sample project supplied shows, does not contain any audio tracks and that the artefact sound is that generated by VSTs used within Shared Layers and played near the transition from one Part to another Part.

Thanks anyway for the cue provided! :blush:

1 Like
18

Hi @musicullum,

over the past few weeks, I have spent a lot of time trying to definitively understand the terms of the problem that has been exposed for months in this post and in some previous ones concerning the issue.

I truly believe I have come to a firm conclusion, and this is supported by the realisation of some workarounds made to try and ā€œbypassā€ the issue and continue using the projects I have in place.

First of all, there seem to be two types of problems:

  • Systematic generation of artefact sound

  • Occasional loss of processing of a MIDI message

Both of the above-mentioned issues are related to the use of Parts and manifest themselves when in the timeline there is a transition from one Part to the next Part.

The generation of the artefact sound seems to be exclusively inherent to the use of the Shared Layer and is, as already written, influenced by the value of the ASIO Buffer Size.

The occasional loss of MIDI message (note) processing, on the other hand, seems to affect Layers regardless of type.

The conditions necessary for the loss of the MIDI message to occur seem to be many: note played sufficiently close to the beginning of a Part, the complexity of the VL project, probably also the computational power of the hardware used.

One thingā€™s for sure: all I had to do was to take one of my projects, specifically a Song in which the drummer, who uses an e-drums, was systematically encountering the issue of lost notes, set back all the triggers defined in the Song and associated with the Parts by a quarter beat, and the issue ā€œdisappearedā€! :open_mouth:

Obviously what happened was that when the drummer was playing the note at the beginning of the Part (which typically falls at the beginning of the measure or at the beginning of the beat), having set the triggers back determined that the Part had already been reached for a ā€˜whileā€™ and that it was therefore far from the beginning.

Itā€™s still unclear to me whether you were able to reproduce the issue and aware of this problem, I hope you can finally tell me something about it. :pensive:

I am hoping for your feedback that will shed some light. :face_holding_back_tears:

1 Like
19

Your construction is somewhat adventurous :slight_smile:

But we figured out what happens; when the Part changes, there is still notes playing, and thus de-activation of the Layer is postponed. Then when all is settled, it is disabled which causes an all notes off to be sent. However as that plugin is shared, it will cut off the current operation. Should be fixed with the next prerelease (tomorrow, if all goes well), check again then, thanks.

4 Likes
20

Hi @musicullum,

obviously, my assumptions are the result of observing a ā€œblack boxā€ of which I only observe the external behavior and for this reason they can sometimes be a bit ā€œadventurousā€. :slight_smile:

I think I understand the scenario for which the new fix was prepared, but unfortunately it doesnā€™t seem to coincide with some of the tests I have carried out.

Of course, Iā€™ll test the new version and report back to you.

As always, thank you for all your support! :blush:

1 Like