Quarta-feira 7 Jan 2009, 13h:39

I would like to share with you some details about OMR generation, the possibilities, and possible future enhancements.

For now, the algorithm uses combined: standard deviation of tags percentage, tags number and a few constants.

The script allows the use of: variance/standard deviation of top 1-x tags, variance/standard deviation of top 1-500 artists, we have a set of data for each artist like the number of tags, the percentage distribution of each tag, the number of plays…

Any suggestions on OMR improvements using the mentioned data are highly welcome :)