Converting midi to audio timing issues

Hi to all,
I have some issues with a conversion from a vst plugins to an audio file: the result audio is not aligned perfectly to the grid and is in advance related to the beat.

The steps to reproduce this problem are:

  1. add a vst instrument to an empty project (in the attachment you can see I’ve used Synthmaster in the 1st track)
  2. convert the vst to an audio file. I’ve used two techniques: the first is to add a group channel (track nr. 2 in the image), route the audio from the vst to this channel and then route the group channel to an audio track (in the image the track nr. 3, called “Audio 01”).
    The second try was to bounce the audio (using file-export-audio mixdown) and reimport it in the project.

As you can see in the attachments both channels “Audio 01” (track nr. 3) and “synth master” (track nr. 4) have some issues with timing. If I zoom in the wave form of both these tracks the audio is starting in advance. Is this a normal behaviour? Is something related to a setting in Cubase?

I’ve tried with different vst instrument, but the result is the same.

Thank you for any help.

Luca
gaps.png

Hi,

This is normal for all DAW (actually for all digital audio devices). You even don’t have to use different export techniques. If you export the same signal twice from the VSTi, it’s always slightly different. Even if you play it.

The reason is ASIO Buffer. The Buffer is always in slightly different position, so the oscillators start in a different position, related to the Ruler.

Hi Martin thank you for the reply. I didn’t know about that, so I think I can ignore the small gap.

Luca

One thing I’ve noticed: I get better results if I freeze the midi track and then do an audio mixdown. All the exports are consistent and the gaps are very small.

Yes, because the “export” is done while freezing, actually. But, if you try to freeze multiple-times, and save all frozen files, you will see exactly the same result. Every frozen file will be slightly different.

The same is for the Render In Place.

But do you think does it worth to warp the exported audio to have the best timing? Or the gap is so small that it doesn’t make any difference?

A few thoughts if this is really bugging you (especially if you can hear a timing issue)…

I typically move the track event as necessary to “line up” with the other tracks. I do it this way because I like to have track timing aligned visually. So this my preferred method (I know I’m weird :wink: ).

But… a more “civilized” way to fix the timing is to use the “Track Delay in Milliseconds” function. This function is available in the inspector 1st tab. The field looks like a clock with arrows on each side pointing left and right. You can adjust the timing either way up to (I believe) 2 seconds.

Note: If you know you are going to have this issue before rendering or mixdown you can also adjust the timing on the instrument or midi track before rendering (or mixdown) to audio. The resultant rendered audio track will be lined up with other tracks and the “Track Delay in Milliseconds” will be set to 0.00.

Regards :sunglasses:

You will get exactly the same timing, as if you export it. I mean, the same range of various. When you will be lucky, then you will be closer to the original.

What should affect it is the Buffer Size. Lower Buffer Size (while export) = lower time range.

If you want to make it really super-precise, I would recommend you to use Render In place, and then shift the exported Audio event, and align it manually.

But honestly, do you hear it? I’m not using 100% quantize, and I count with different attacks of the sound, so in my music, it doesn’t fit so tight, to keep it a bit live.

Ok got it! I agree with both of you. I can adjust the timing with track delay but at the end we are talking of 0.1 ms in the worst case. I think no one on this earth would hear such a tiny delay.

I thought computer music was somehow too much precise, but I was wrong :slight_smile:

Thank you!

Actually, I wouldn’t use the Track Delay here. Only if you would need to quantize the signal later on, for some reason. But I expect, it’s the final track, already. I would use the Track Delay, because you don’t have a visual feedback on it, so ti would be still shifted visually, what is a bit annoying, if the goal is to align it.