I mainly use spark to film short interviews. When cutting around an interviewee's statements, sometimes you need the audiovisual cut to be EXACT. It is frustrating that you cannot fine-tune the frame where the cut happens, but what would be helpful is if the preview video audio cuts were EXACTLY the same as what is ultimately downloaded. So many times I have exported only to find that Spark has "guessed" where I want my a/v cut to be when generating the preview, and only "cleaned it up" in the export. The result is that I cannot rely on the preview to accurately convey the video as I am envisioning it, and instead have to repeatedly download the video in order to be assured of the timing of my a/v cuts.