Be aware of fluctuating Instrument Track Timing when layering with audio events

[FMN-Music]

This is another classic one I nearly forgot about, but came across again today.
I wanted to turn Drum Sample Instruments to Audio to save cpu and get project latency down, but recognized differences in sound afterwards, so I checked with polarity reverse and was reminded of the fact, that Instrument Track Timing in FSP is somehow dependend on the Playback starting point.

So be aware, that layering Drumsounds from VST-Instruments with Drumsounds from Audio Events will give slightly different results, depending on where you start your playback. Especially in a more complex song. It's a weird one.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

Also be aware that even in simple songs, the timing between VST-Instruments and Audio Tracks is flakey*. Severity seems to be influenced by the song tempo. Might be an issue with PPQN Resolution (Pulses per Quarter Note). If I remember correctly, PPQN Resolution (or Ticks?) was offered as a setting in Cubase.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

* Timing between VST Instruments themselves seems solid, as well as timing between Audio Tracks themselves.

 

[FMN-Music]

Found my original video on this topic.. from 7 years ago.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[Skap]

 

[Skap]

Thanks for bringing this up. You sure have a talent when it comes to finding and keeping track of these issues. Wish the developers contacted you and had a conversation about this and the other timing related bugs.

(These bugs and shortcomings make me uncomfortable. Vibe killers. I remember going down the rabbit hole of automation timing a few years ago on the old forum, and I found that different buffer sizes would produce different timings of the automation when rendered to audio.)

 

[Rambler]

Is this a Studio One/Studio Pro specific issue or is this simply explained in the difference of accuracy between midi ticks and samples?

 

[Rambler]

Just tested this in Reaper as well (3 virtual instrument drum tracks with processing, rendered to audio with a flipped polarity on those audio tracks)

Interestingly, the result is very similar.

- If I start playback on 0:00:00 -> perfect nulling, -inf on the Master Bus
- If I start playback somewhere in between -> does not null anymore, different results depending on the playback postition

Maybe this is inevitable due to MIDI's less precise timing compared to audio? Or Is there something in S1 that differs from other DAWs in this regard? Anyway something to keep in mind when working with MIDI and audio in combination (layering, drum replacement etc.)

EDIT: Interestingly enough, in my test here S1 doesn't do a perfect nulling starting at 0:00:00 as opposed to Reaper.

 

[FMN-Music]

Interesting, in my test I could not get Reaper to fall over.
But I can't say that this timing issue is an exclusive FSP "feature", because Ableton 12 exhibits the same problem. Works at 120 BPM, but does not at other tempi.
Cubase (like Reaper) performs this task flawless. The Tick setting in Cubase does not affect the outcome.

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[Rambler]

My test included different tempos.
Turns out the result is actually depending on the tempo + start position.
Interestingly 120bpm and 250bpm always seems to null out independent of the start position (apart from one pop when you press play).
With the other tempos (e.g. 132bpm) it doesn't null out completely and it seems to be dependent of the start position. I gotta say though that the differences are small enough to not be audible, I can see them on the meter)

I would include different tempos in the test and see if that changes results
(Edit: I saw you did change the tempo in Reaper but I would try more different tempos and start positions)

 

[BobF]

How does [Chase long notes] affect these tests (if at all)?

 

[FMN-Music]

If the cursor is positioned inside the boundries of a midi note, it affects this first midi note at playback start, but should not affect following Midi Notes.

Midi Note Length is also a factor not to be overlooked, when changing the tempo. Midi Notes get shorter at higher BPM (audio events do not) and therefore the Sampler might not play the whole sample anymore, resulting in Null Tests failing.

 

[Rambler]

If the cursor is positioned inside the boundries of a midi note, it affects this first midi note at playback start, but should not affect following Midi Notes.

Midi Note Length is also a factor not to be overlooked, when changing the tempo. Midi Notes get shorter at higher BPM (audio events do not) and therefore the Sampler might not play the whole sample anymore, resulting in Null Tests failing.

To rule out anything the sampler does, for test purposes it’s probably the best to render the audio file from the instrument tracks and having the BPMs set before doing that

 

[BobF]

If the cursor is positioned inside the boundries of a midi note, it affects this first midi note at playback start, but should not affect following Midi Notes.

Midi Note Length is also a factor not to be overlooked, when changing the tempo. Midi Notes get shorter at higher BPM (audio events do not) and therefore the Sampler might not play the whole sample anymore, resulting in Null Tests failing.

I agree that it should not, but does it in practice?

 

[FMN-Music]

Tried it, did not change anything. Midi Notes in my tests were not even long enough to be affected by this setting at all.