Sounds sensible. Seems like there are a few .net wrappers for sqlite. Could be advantageous to move the query so that it's network-based (off the top of my head, a web query) rather than off a file that you have to distribute with the player?
Yes. I'd been planning to add support for pulling data from the network. I actually registered openriff.com at some point, with vague intents to set up a web backend for this purpose.
While I was working on my player, I was wondering a bit about using fourier transformed snippets from the feature audio to serve as guides or markers for the rifftracks. That way you wouldn't have to make new offsets (and potentially rates) for every edition or rendition of a feature (or make assumptions about its chapter structure). Of course, it wouldn't work with dubbed features. Probably a lot of work and tweaking, too. I'm also not sure about the resolution (which would affect the file size of the guides) needed to make it sync up properly.
Earlier in the thread we'd been talking about fingerprinting the video data, but I think you may be onto something with using the original audio track. The audio waveform should be much less affected by encoding, and I'd think it would be easier to work with than trying to fingerprint the decoded video frames.
