Warped Files Bounce Inaccurately

[alexis]

@Grim

So every audio export is identical and correct.

Your test proves that the 2 exports are identical…but not really that they are correct.

So in your test after step 3, in the project window do your waveforms line up?..once zoomed in with my own test I am finding various hits within the file to be anything from 20-123 samples different from the original (or the appearance of).
I did export both realtime & offline & these do line up perfectly but both are way out to the original warped waveform.

Reading between the lines here after the Steinberg response, I would say that the export is correct but the warp algorithm is way less accurate than it needs to be and so the waveform view does not reflect where warped hits will play.

Nice pic. I can confirm on my system several things I observe from yours:

1. The original warp is out of time w/ the bounced warp.
2. The original warp is out of time w/ the exported/imported warp.
3. The original warp has different data than the bounced warp (compare peaks and valleys to prove that).
4. The original warp has different data than the exported/imported warp (“–”).
5. The bounced, and exported/imported, warp are at different volumes than the original warp.

[Grim]

Just to clarify my files were both export audio mixdown…one realtime one offline.
Didn’t try bounce selection though I would expect the same result.

3. The original warp has different data than the bounced warp (compare peaks and valleys to prove that).
4. The bounced, and exported/imported, warp are at different volumes than the original warp

Yep…the timing and the shaping of the waveforms are all over the place. Here is a short fill where I lined back up the first hit to the original (top)…look at how far out the 2nd and 3rd hits are and how softened the transients are.
Seems like the algo struggles to keep up when warp tabs are closer together.
Actually the peak level of the export is louder than the original despite most of it looking quieter.

[mr.roos]

Wow, a Steinberg response, great! But yeah, like Grim states, it’s not the comparison of the exported copy to another exported copy, but instead the comparison of the original warped track to the bounced or exported track. But did I misunderstand your example?

Alexis, I see the nature of your question but try this as I described. I seriously doubt the warped track that I copied became altered. No, I had not heard this about the copy process nor have I seen this happen. I guess the only test for this is is to make an external recording of the warped track on playback and then copy it and see what you get. Pretty convoluted hoop there for something I doubt happens. As I think about it, that a copy would be inaccurate against a noncopied original, well, this is buggers.

Hm. Here’s a test. Do this in this order:

1. warp a track
2. bounce it and compare it
3. copy the original warped track, compare it to the original
4. compare all three versions

My thinking is that if you can recognize the differences in the original vs. bounced, then you could see if that relationship changed after you copied it.

Pristudio, hm, I don’t understand how it is that you were trying to bounce a warped track, and you could use that; but now you can’t use a copied track. Wha?

[Marcus]

Hi,

following your steps and just one export, then reversed the phase, export again and the results are in my attached image.

As you can see the result shows just some fragments / noise. This seems to be related to the algorithm.
As I have written before, we will improve the Algorithm (Warp Tool) in the future.


Cheers,

Marcus