I see. The next step then is a debug log from people experiencing it (ideally attached as part of a GitHub issue comment, but since not everyone has a GitHub account...).
Can people try doing a data analysis based on information we get from users - NVDA versions in use, debug log analysis, etc.?
