Jarring audio "pop" when using Voice Isolation

Hi all,

So this happens intermittently, but it's been happening to me since Voice Isolation was introduced, and I'm still having the same problem in 18.6.6.

I have a number of quick cuts (1-2s) in succession that all have Voice Isolation applied.

I've put a short (0.3-0.5s) cross fade between them.

They're in a "scene" timeline that is then included/nested in a main timeline.

1. I have to cache the effect for it to work properly on the main timeline, otherwise I get a weird "echoy" effect on the final render. You can hear this during preview.

2. Most problematic, every so often, the final render has a huge "pop" on the transition. You can't hear this in preview, only on the final render.

And it seems completely random, so the pop might end up on a different transition each time you do a final render.

I know there are a lot of issues with nested timelines, but I'd prefer to keep that workflow if possible.

Has anyone else had a similar issue?

(edit) Sounds like this has been a well-known problem since ~2021 or so...
viewtopic.php?t=140926&p=759305

dr-1.jpg (17.93 KiB) Viewed 537 times

dr-2.jpg (3.05 KiB) Viewed 537 times

dr-3.jpg (10.23 KiB) Viewed 537 times

I know there are a lot of issues with nested timelines, but I'd prefer to keep that workflow if possible.

It may not be.

They're in a "scene" timeline that is then included/nested in a main timeline.
And it seems completely random, so the pop might end up on a different transition each time you do a final render. I know there are a lot of issues with nested timelines, but I'd prefer to keep that workflow if possible.

Yeah, I'd say don't do that. My guess is that the system is getting stressed out and can't handle processing all the video AND the audio at the same time.

A possible fix: render out all the picture as a flattened file to a mezzanine format (like ProRes 444 or DNxHR444). Create a duplicate timeline and delete all the video, then bring in the flattened mezzanine file (with no color corrections), and see if that will now export without any audio glitches.

As Jim commented, I would suggest avoiding nested timelines, particularly when heavy audio processing is involved. For large projects, we do all the color on one system and then hand over all the sound files to a local sound mixing house, they do the entire project in something like Pro Tools, then they export out all the 5.1 tracks and send those to us. We bring those tracks into Resolve and do the final deliverables there. You could do something like that in Resolve, just in multiple steps.

Thanks, yes, that's what I assumed. I'm trying out a workflow where I bounce all the audio to a new track and then mute the original tracks. This is showing some promise, but I'm not ready to celebrate yet.