Why is bounce in place not sample accurate?

[stash98]

Ok, after some more tests:

Bounce if broken when you adjust the size of a clip. If you adjust the clip start time to make it shorter, then move back the clip to the start of the bar and then bounce, it will be bounced later than the original file. I tried consolidating the clip and doing a bounce in place, but the issue persists.

EDIT: I read through this thread again and OP seems to have had the issue with clips that had already been chopped or resized. So, if you import a wav file and bounce it without altering it, everything works fine. But do anything to the clip and the bounce will be out of place.

Thank you for that. So if I recorded some long midi takes on multiple channels and did not adjust them at all, I could bounce and the audio would be sample accurate? If so I can at least work around with that.

[Danilo Villanova]

Ok, after some more tests:

Bounce if broken when you adjust the size of a clip. If you adjust the clip start time to make it shorter, then move back the clip to the start of the bar and then bounce, it will be bounced later than the original file. I tried consolidating the clip and doing a bounce in place, but the issue persists.

EDIT: I read through this thread again and OP seems to have had the issue with clips that had already been chopped or resized. So, if you import a wav file and bounce it without altering it, everything works fine. But do anything to the clip and the bounce will be out of place.

Thank you for that. So if I recorded some long midi takes on multiple channels and did not adjust them at all, I could bounce and the audio would be sample accurate? If so I can at least work around with that.

Sorry, I didn't do any testing with MIDI.

[stash98]

Ok, after some more tests:

Bounce if broken when you adjust the size of a clip. If you adjust the clip start time to make it shorter, then move back the clip to the start of the bar and then bounce, it will be bounced later than the original file. I tried consolidating the clip and doing a bounce in place, but the issue persists.

EDIT: I read through this thread again and OP seems to have had the issue with clips that had already been chopped or resized. So, if you import a wav file and bounce it without altering it, everything works fine. But do anything to the clip and the bounce will be out of place.

Thank you for that. So if I recorded some long midi takes on multiple channels and did not adjust them at all, I could bounce and the audio would be sample accurate? If so I can at least work around with that.

Sorry, I didn't do any testing with MIDI.

All good, I will test this out soon and report back in this thread.

[Danilo Villanova]

Ok, after some more tests:

Bounce if broken when you adjust the size of a clip. If you adjust the clip start time to make it shorter, then move back the clip to the start of the bar and then bounce, it will be bounced later than the original file. I tried consolidating the clip and doing a bounce in place, but the issue persists.

EDIT: I read through this thread again and OP seems to have had the issue with clips that had already been chopped or resized. So, if you import a wav file and bounce it without altering it, everything works fine. But do anything to the clip and the bounce will be out of place.

Thank you for that. So if I recorded some long midi takes on multiple channels and did not adjust them at all, I could bounce and the audio would be sample accurate? If so I can at least work around with that.

Sorry, I didn't do any testing with MIDI.

All good, I will test this out soon and report back in this thread.

Please do. If you have time try to test with rendering and re-importing stems. I suspect there might be an issue there.

I did substantial tests with Studio One 5 and Reaper 7. Studio One is a mess. It doesn't compensate for PDC with TB EQ 4 at all. Even with all eq bands at zero. Reaper is absolutely solid, maintaining sample accuracy even with heavy EQ curves.

[Danilo Villanova]

Anyone know if this is fixed? I went back to REAPER after I found out about it. Pretty insane bug.

[synthique]

There's no way to test this with MIDI because there's no source audio to compare with the bounce. This is a rabbit hole that doesn't have anything interesting at the bottom of it.
Tone Booster's EQ 4 algorithm is non-linear, so there's going to be some frequency shifting going on.

[ihearanewworld]

Anyone know if this is fixed? I went back to REAPER after I found out about it. Pretty insane bug.

You left a DAW because of a one sample shift? How was it negatively affecting your music?

[Danilo Villanova]

Anyone know if this is fixed? I went back to REAPER after I found out about it. Pretty insane bug.

You left a DAW because of a one sample shift? How was it negatively affecting your music?

It's not one sample. Where did you get that from?

[fwsuperhero]

Anyone know if this is fixed? I went back to REAPER after I found out about it. Pretty insane bug.

You left a DAW because of a one sample shift? How was it negatively affecting your music?

It's not one sample. Where did you get that from?

Can you send a reproducible example where you get >1 sample diff?

[nowiamone]

I wasn't even aware of how far back this issue dates.
I tested Bitwig 4.4.10, Bitwig 5.1.9, and Bitwig 5.2 beta 7, all have issues with bouncing audio samples to audio under specific circumstances (bitrate/stretchmode/rec-arm of Track - dependent).
Here's a video which shows some of these findings. Based on the video, you might be able to reproduce these bugs. I already sent that video to Bitwig of course (i basically made it FOR Bitwig . I really really hope they take the time to fix these bugs once and for all. It's mind boggling to realize this has been present for several years already. With Beta 7 it even got worse, because the first transient gets drastically cut off on Tracks which have rec-arm enabled

PS: I made a little mistake when comparing 5.1.9 with beta7, not having the same stretchmode for short samples active, but nvm, the results are still similar / the bugs (except for the first transient reduction) still exist in 5.1.9

[jpcaulfield]

From Bitwig Support :

" Sometimes 3rd party plug-ins report back an incorrect latency, in this case there will be problems with the timing of bounce in place. Are you able to reproduce the issue without any 3rd party plug-ins involved? "

This is very frustrating , but also common in Ableton etc. It's most deadly with Kick and Bass, I would manually fix one kick to time marker and duplicate it and the same with bass and maybe even hi hats. an excessive amount of extra work .

[nowiamone]

From Bitwig Support :

" Sometimes 3rd party plug-ins report back an incorrect latency, in this case there will be problems with the timing of bounce in place. Are you able to reproduce the issue without any 3rd party plug-ins involved? "

This is very frustrating , but also common in Ableton etc. It's most deadly with Kick and Bass, I would manually fix one kick to time marker and duplicate it and the same with bass and maybe even hi hats. an excessive amount of extra work .

Good question though. Does it happen with Bitwig internal devices?

But in general, bounce in place is sample-inaccurate, independently from any plugin. Even when bouncing AUDIO. haha. So support seems to ignore what i presented them a while ago.

[liquidsound]

Get this, on Bounce Serum is almost Sample accurate while Polysynth is not...

Using the Time Shift (Just to see a difference):
PolySynth needs -37 Time Shift Samples
Serum only -3 Time Shift Samples. (edit)

That's just a comparison using the Time Shift device. Nothing more.

[luzid]

Bouncing a raw audio clip and inverting the polarity doesn't even sum to zero. The master channel shows a level of -157dB. Bouncing an audio clip repeatedly increases the level of the resulting mix. After 10 iterations it's at -130dB.
Out of curiosity i boosted the mix and it sounds like noisy crackling.
Apparently bouncing introduces some kind of quantization noise. I'm bouncing 32-bit without dithering.

The same experiment in Reaper works as expected. Zero error or noise.

[nowiamone]

Bouncing a raw audio clip and inverting the polarity doesn't even sum to zero. The master channel shows a level of -157dB. Bouncing an audio clip repeatedly increases the level of the resulting mix. After 10 iterations it's at -130dB.
Out of curiosity i boosted the mix and it sounds like noisy crackling.
Apparently bouncing introduces some kind of quantization noise. I'm bouncing 32-bit without dithering.

The same experiment in Reaper works as expected. Zero error or noise.

Please share these findings with Bitwig. They got to BELIEVE that Bouncing/Bounce in Place has issues which need to be fixed...

I especially hate how Midi-to-Audio has it's issues, too. Because in this case, "Proving" the issues is much much harder than showing issues with Audio-to-Audio.