Like many we deal with interviews and, unlike many, wired athletes. We want to archive this media to LTO but data granularity is a problem with this content. In both cases cameramen just keep on rolling creating very large files. If a short segment from these long files is needed from the LTO archive, then the entire file must be restored. Our solution to this is to scene detect the media and subclip prior to archiving. Unfortunately interviews don't scene detect because the scene never changes. Using audio silence or volume differences as scene criteria would allow this media to be "scene" detected. It would need to include a high to low volume only break point so when the interviewee was done, it would break. This would allow the next question to be included and not break when volume increased on the answer.
Another scene detect feature request... a minimum scene length setting.