There are a number of features and applications that need to know where sounds are in the middle of MP3 files. For example:
- Audio search finds words and phrases in audio files and keeps track of where they are.
- Our own Clip feature likewise works on the basis of offsets by time.
These are great features, but they’re plaugued by another new development: they dynamic assembly of MP3 files to include promos, commercials, etc. The problem is that when MP3s are re-assembled with components of different lengths, any previously identified sections will now be earlier or later in the file than previously determined. In other words, if a show is reassembled and therefore made longer or shorter, the location of a clip or segment will move and the search engine or clip feature will return incorrect results.
I’ve been working on a solution to this problem, and although it isn’t perfect yet — and maybe never will be — I want to get it out there to see if others would be willing to support it. We need buy-in from those who assemble shows or insert spots, and we need buy-in from the audio search engines as well.
My idea is to place into an ID3 frame information that will allow search engines and clip processors to understand what has been inserted or removed from the file so that these tools can deliver consistent results. Some possibilities include:
- ID3 “Event Timing Codes” (ETCO frames)
- ID3 “Position Synchronisation” Frame (POSS frames)
- A custom ID3 frame
I haven’t figured out how POSS frames work, so they may have no relevance whatsoever. But one way or another, my idea is to somehow put offset information into ID3 frames, which can then be updated when the position of components within an MP3 file are changed.