Did anyone ever find an answer to this question? I have found a few posts with similar questions which have also not been resolved. I have read the XMP specification documents (specifically part 2) and I still can't piece together how to convert the starttime and duration tags into milliseconds.
XMP stores time values in a unit called "ticks". The conversion factor is 254,016,000,000,000 per second. This value was chosen so that we could have integer representations of all times in all of the common frame rates.
Thanks Todd. See snippet of code below:
The word "what" occurs at 13 milliseconds. I say that because when I run speech to text analysis on the embedded astx file, i can then click on the word in Premiere. The word is associated with a time code of 00:00:00:13. If I divide the startTime by the frameRate (114053184000 / 254016000000), I get .449 which doesn't seem to jive with 13 milliseconds. Is my approach wrong in some way? David