Based off the okay Samurai Dec 2016 Tips & Tricks video, that was excellent. It works excellently even recording wide and tight parent groups with audio being recorded off the mic. When I use the "Compute lip sync from scene audio" the lips only move on the default parent group even though on recording, I kept changing triggering between my tight and wide parent groups. The final recorded video shows only the lips moving on my default- the wide, but while the head movement records on both, the lips don't move on the tight. I want to add my backgrounds in AE but need the wide and tight shots in one take so that the body/face/eyes gaze are uniform between the shots. Is there a way around this?
I just tried this with the Red Monster Live setup (files available in the video description at https://youtu.be/qBUOKI22hzI ) and it's working for all camera views with computed scene audio, so I must have done something that makes it work here. Take a look at that file and see if your setup is different. If I do the scene audio computation, I see three "Lip Sync" tracks appear below my main "Red Monster Live" track.
Worst case scenario, could you share here or DM me your .puppet file (File > Export > Puppet) via a shared link on Google Drive, Dropbox, or Creative Cloud to take a closer look?