After doing all of the above:
1) Reverting to 2016.12
2) Bouncing all of the edited clips to make a single clip per track with no splits
3) Cloning my vocal track to a whole new track that had none of the clips or events, just the effects stack, and then dragging in the audio
4) Creating a whole new automation envelope from scratch
it played correctly in real time under SPLAT. But then I was back to exactly the same problem I originally had, from the first picture above. Where I pull down the fader after the first vocal section, the vocal track disappears a few seconds too early in the WAV file. GRRRRRRR.
There are serious problems here. The WAV file should match what you hear in real time playback.
I note two other things that might contribute to the problem:
a) My clips all start around "measure 8". This was a live recording that bears no relationship to the SONAR measures, and the automation envelope should not depend on the starting point.
b) I did extend the start and end times on the ruler by a few seconds to make sure I didn't have residual garbage in the front and to make sure I got the full tail at the end. Again, that "should" not affect the envelope, but I guess I will have to test both of those conditions now.
If I bounce everything (entire mix) to a new track, I get exactly the same problem as in the WAV -- the vocal cuts out early.
====
On edit, I moved all the clips completely to the left (measure 0). The real time playback continues to be perfect and the exported WAV continues to have the vocal cut out about 3 seconds early.
post edited by cparmerlee - 2017/02/08 15:51:53