The duration of this whole track is actually 25:06 min
I think you're on to something though, I'm thinking that number is somehow related to the frame/bit rate. These XMLs are meant for tagging video at a particular point in time (relative to audio). If you have any further insight as to the formula I would use to deduce the actual point in time (preferably in units of seconds/milliseconds/minutes) that would be awesome.
okay so...from developer documentation, this only partially answers my question but a start:
"Time values are expressed as a rational number of seconds with a 64-bit numerator and a 32-bit denominator. Frame rates for NTSC-compatible media, for example, use a frame duration of "1001/30000s" (29.97 fps) or "1001/60000s" (59.94 fps). If a time value is equal to a whole number of seconds, the fraction may be reduced into whole seconds (for example, “5s”)."
I found this by googling "fcpxml audio tag" and clicking on the pdf
Being a video edit XML I imagine that there is a formula based on frame rate. I know broadcast wav timecode data is based purely on number of samples from a zero point. I am guessing it is for FCPX as roles are mentioned.
However I have no real skill in this area apart from some involvement in beta testing and software development with some audio companies.
Many years later, maybe it's still worth replying in case others stumble onto this post.
In .fcpxml files, the start, offset and duration numbers are real-time seconds (as hinted to by the "s" at the end). They are not related to framerate. You only need the framerate if you need to convert these times to a frame number or to standard timecode which uses frames.