Out of phase audio on export of sequence with < 1 frame timing adjustments between synced tracks

[videowizard.io]

I sent an audio track to adobe podcast for enhancing. I drop the enhanced track underneath the original (I like to mix the results instead of using just the Adobe Podcast audio).  

The mixed result is out of sync by less than one frame. You hear it as a slight phase shift, or chorus effect. I "show audio time units" and nudge the Adobe Podcast audio so it's perfectly aligned and in-phase with the original audio. Problem solved. Or so I thought. 

On export, the chorus effect returns. It sounds like each audio track is independently 'slipped' back from audio-unit timing back to the more blunt frame-level timing.

Any tips for maintaining "what you hear is what you get" audio from timeline -> export, on sequences with minute, less-than-one-frame syncing?

[mstegner]

Is your source file an MP3?  If it is, it might be the cuase.  

MP3 audio content can be shifted depending on the encoder and decoder used, and there are various different encoded used in the processing of the audio.  This can cause small shifts that are not notable until the processed audio is mixed with the unprocessed audio.  Until this can be addressed if I suggested that the source be converted to an uncompressed format such as wave before uploading to the Adobe Podcast Enhance Speech service.