• 97 Posts
  • 509 Comments
Joined 11 months ago
cake
Cake day: June 24th, 2023

help-circle






  • Other answer seems to suggest that the problem is that the same podcast can be available, depending on where and who is listening to it, with different length due to different ads injected into. Here’s my probably stupid and completely ignorant suggestion: instead of using timestamps for both begin and end of the ads segment, you could use a timestamp for the beginning, and an hash of the first part of “non-ads” segment. I’ll try to explain better:

    |----------------xxxxx--------------------|
                    ^     |___|
    

    The xxx is the ads segment, the ^ is the timestamp of the beginning of the ads, the |___| is a small duration segment (for example, 0.5 seconds) right after the ads segment. The data of that segment is hashed and used as “end ads segment indicator”.

    On the other device, with a different duration of the ads, you should start hashing it to find the corresponding segment.

    Is this doable or did I just said a bunch of idiot things?