Interpretation Out-of-Sync in Webinar Recordings

The org I work for recently completed a multi-day online conference using Zoom webinars, and the entire conference was recorded to cloud. There was also Spanish and ASL interpretation for the entire conference.

As I have been working with the various recording files, I have found that the Spanish interpretation audio is consistently out of sync with the video files. This is occurs even if I give a generous lead gap for the start of the Spanish audio... the interpretation ends several minutes before the video and falls further out of sync (ahead of) the video as it gets longer.

I have discovered that I can approximately correct this discrepancy if I slow-down the interpretation audio by around 5% (playback at 95% speed). The audio at 95% also sounds more natural, which makes me think that this discrepancy is an artifact of Zoom's audio compression.

Has anyone else encountered this discrepancy in their interpreted audio in Zoom recordings recorded to cloud? Does the same problem exist in recordings to desktop?