My premises are:
-Storage is cheap (keep as much info for every file as possible)
-Computing time is cheap (this is a one-time operation)
-The users free-time is expensive
-Good tags should never be replaced by bad ones
Since humans can do it, I have no doubt that it CAN be done computationally. I also guess that our human database is smaller than any FREEDB/CDDB. However, this is touching AI, and I guess that someone with actual experience in that field can crush all my hypothetical ideas? =)
In one perspective, one is doing an inverse lookup of the CD ID. When you are ripping a CD, FREEDB will do a database query based on a unique code on the CD. This gives very solid information. If only that ID can be found for a set of files forming a CD, then obtaining good tags is a piece of cake.
-k
-Storage is cheap (keep as much info for every file as possible)
-Computing time is cheap (this is a one-time operation)
-The users free-time is expensive
-Good tags should never be replaced by bad ones
Since humans can do it, I have no doubt that it CAN be done computationally. I also guess that our human database is smaller than any FREEDB/CDDB. However, this is touching AI, and I guess that someone with actual experience in that field can crush all my hypothetical ideas? =)
In one perspective, one is doing an inverse lookup of the CD ID. When you are ripping a CD, FREEDB will do a database query based on a unique code on the CD. This gives very solid information. If only that ID can be found for a set of files forming a CD, then obtaining good tags is a piece of cake.
-k