SD3 MIDI playback engine phase issues

[John Ohle]

Hi,

I bought SD3 and I was testing it out within my DAW.
I use MIDI quite a bit naturally with my music and before jumping in with SD3 (over SD2) I decided to run a few tests.

I use Logic Pro X (10.2) on OSX 10.10.5 using a mid 2014 rMBP.

In my tests I noticed that if I ran the following sequence that the sound cancelled as expected (phase inverted cancellation):
1 – created a sample MIDI file in the Logic Main Window on a SD3 Instrument Track *,
2 – duplicated the track and all settings*
3 – setting the midi file velocity levels to a static level (127) for all hits,
4 – copy the midi to the duplicated track and then used the “Gain” plug-in to invert the L&R channels in the duplicated track.
5 – no sound will be produced as the tracks will cancel.
*using the default kit and DISABLING ALL of the “Hit Variation” parameters for ALL instruments in the default kit of course.

However, running the same test but copying the exact same MIDI file (with set velocity levels to 127) to run WITHIN SD3* did not 100% phase cancel when the duplicated track was played alongside the original SD3 “internal MIDI” track. It did cancel to some extent but not fully.

As far as I am aware I disabled all of the variables in this test with constance MIDI levels and disabling the “Hit Variation” parameters for ALL of the drums and cymbals in the kit.
Additionally, the phase cancels correctly when the MIDI files are in Logics Main Window.

Why does it matter when the MIDI file is played back within SD3 and not in Logic?
Why does the behaviour here go against what I would expect to happen within any DAW? i.e., the produced audio from SD3 should be identical with this scenario though it seems that SD3 is not behaving as expected.

Is it a bug in the MIDI playback engine within SD3?

I have attached my test MIDI file for anyone who wishes to try to reproduce this and a screenshot of my project to give you an idea of my project (I tried to upload the entire Logic project file itself but it is too big for this forum).

Thanks!

Topic

[Olof Westman]

What sample rate are you running?

Olof Westman - Toontrack
Coder

[John Ohle]

I’m 99.9% sure it is at 44.1KHz by default in my Logic settings.

Anyways, it wouldnt matter surely as with Logic running the MIDI, duplicated and inverted, cancels out correctly.

If SD3 was using a different sample rate than Logic then the playback would sound dodgy but it doesnt.
On its own the sound quality is fine, but it does seem to have some odd phasing issues as I reported above.

You can try the reproduction steps I mentioned in my Original Post to see for yourself. 🙂

[Olof Westman]

>Anyways, it wouldnt matter surely as with Logic running the MIDI…

Err… maybe. SD3 runs at 44.1 internally, always. As soon as you run another
sample rate the incoming MIDI has to be converted to 44.1 samples. I should
think that there is a risk of some difference in rounding placing notes on adjacent
samples rather than on the same sample, occasionally.

However, there would seem to be some other issue here. Thanks for
finding out!

Olof Westman - Toontrack
Coder

[John Ohle]

Sample rate with Logic “playing” the midi files results in 100% cancellation. However with SD3 playing the files, and inverting the output from a duplicate track, as mentioned it does *NOT* result in a 100% cancellation. This suggests to me either:
1 – The audio produced is not in the same phase within each instance of SD3.
2 – The audio produced by SD3 is not the same, even with the “Hit Variance” disabled.
Perhaps this is an issue with some or all of the the Hit Variance parameters not being disabled correctly from the SD3 GUI?
Or maybe its an issue within the SD3 playback engine?

@Olle said:
However, there would seem to be some other issue here. Thanks for
finding out!  

No problem Olle, always glad to help TT out when I can. 🙂
I would hope to see this issue resolved at some point.
While of course, no one runs with a duplicate track inverted in the real world, phase issues can be an issue and within a track it is good to be able to trust that SD3 is not causing any phase issues. God knows I’ve done my fair share of dealing with drum mics out of phase! 🙂

[John Ohle]

Hey,

Any update on this issue?
Was any one from Toontrack able to replicate this issue? If so, is there any fix in the works?

I am very hesitant to proceed with using the SD3 midi editor/engine at the moment as I am afraid that it could cause some phase/timing/other issues within my mixes.
Thanks 🙂

[Olof Westman]

From what I remember we found two issues that both causes problems like those you reported. One was a bug in the SD3 MIDI track that caused some jitter when playing up the MIDI. This has been fixed. The other is harder to get to 100% grips with. It is that each individual MIDI file has a certain resolution in time called “division” and since you mix and match lots of MIDI loops in our song track the MIDI files we create when you export it will have some division that may be different compared to that of the constituting MIDI files. We have increased the division we use when exporting.

So your report led to two changes and they are currently being tested and are likely to be there in the next release, 3.0.3.

On the other hand. You should not worry that these small timing deviations in 3.0.2 should lead to problems with your mixes. They will not cause phasing issues. They will cause very small differences in the timing of the sounds played but thay are so small that nobody before you has noticed them.

Olof Westman - Toontrack
Coder

[Author]

Replies