I tested with Retrologue. Triangle, one oscillator, fixed phase at 0°.
Render in place with dry settings and 32bit float (Cubase’s audio engine runs on 32bit float as well).
The signals null each other perfectly.
I also tested with GA SE 6 and a kick drum sample. Did not do anything in GA SE besides loading the sample.
On every trigger of the kick drum sample I get a very short signal spike. The meter in the right measures it at -138 dB. If I amplify the signal by 138 dB I can hear a certain part of the kick drum sample.
I am not sure why GA SE 6 does not null perfectly.
Thanks for testing. Could you test my latest project, because the spikes are way higher in there. You ned to loop it. I don’t need to amplify even anything. And those spikes will be random. I will test Halion tomorrow.
No, I checked that. The problem is not with the rendered files. I can make two renders from Groove Agent SE (not the render from the render) and those renders will null to each other. But they will never null with the Raw GrooveAvent, because it slightly drifts in time for unknown reason. Those non-null hits will be at random points. Just loop the last project I did. It does show this everything
But situation is even worse than that. You can you SamplerTrack. And it will drift in time too. Make sure to use something like dirty kicks that hitting continuously for four bars long. Then phase invert and loop it constantly. Btw, try to give a single kick a separate midi part and render as “separate events”.
Finally, the Steinberg support answered to this. And yeah, I was right, the issue was real (I’m not sure how did you do your test to not find it….):
“We have reviewed the behavior together with the development team.
The developers were already aware of this behavior, and your tests confirm the current internal understanding.
More specifically:
during playback, a minimal jitter of approximately 1–2 samples may occur;
the file generated via Render in Place / Export can be delayed by 1 sample compared to HALion Sonic 7 and Groove Agent 6 SE playback;
if you move the rendered file 1–2 samples earlier and perform a null test (phase inversion), the signal becomes silent.
This behavior is related to the architecture of the audio engine and its internal rendering process. Since the effect can only be detected through a null test and is not perceptible in a real-world or musical listening context, a corrective intervention is currently not considered appropriate.”
P.S: There are other audio quality issues I found since I started using Cubase (what a surprise, as it is usually considered as very very professional DAW), but I see it is not that important to forum users and Steinberg anyway. Will continue in Bitwig, as it has a way better audio engine. Though I preferred the Cubase workflow more than the Bitwig’s one.
Everytime you render something thats oscilating or modulating, you will not get the same result.
You will have a better chance winng the lottery 10 times in a row or being struck by lightning 7 times than getting an exact render from a synth that oscillates and modulates.
Could you get a guitar player to play the exact same guitar part 100%, this is exactly the same thing.
This is why rendering to audio from midi is important, otherwise everytime you hit play, your synths produce a slightly different set of soundwaves, audio is a pure locked in exact imprint of a moment in time.
Hi. The talk was about static sounds. Completely static. You can read the post above. And Steinberg HAS already confirmed the behaviour which happens due its audio engine internals. And I have various DAWs to test this out. I’m not a newbie.