Hiya!
I'm using FaceFX to generate facial animation for films rather than games, and as such I'd like to be able to use it with long takes - 3 minutes or more of audio.
However, when I run that audio through the default analyzer, it seems to get very confused: mouths open well before lines start, the sync seems wildly off, etc.
Currently I've been cutting them up into smaller takes and analyzing a line at a time, but that's very slow and really causes a hold-up in my workflow. In addition, it would be great to get eye blinks etc in the non-speaking parts of the animation too!
Is there a better way of doing long takes as a single FaceFX file?
For longer files, you should use audio chunk tags (see: http://facefx.com/documentation/2015/W106) If you mark up your text into smallish chunks, the analysis will go well.
As for the non-speaking parts, there should be blinks and such happening in those silences as long as they are part of the audio you're analyzing. However, if the silences are very long, that may confuse the system and cause it to stop inserting events.
Try the audio chunk tags and see how it goes from there.