View Single Post
Old 9th July 2016
  #6
Lives for gear
 

Thanks for the nice comments!

The version shown in the video was a pre-release.

The final version is a little more streamlined. Instead of asking the user to choose among the 7 pre-processing options up front, we are processing all of the options and then we let the user audition each of the 7 results. In practice, most of the options will return the same result. So if 5 of the 7 options result in the same combination of polarity flips, then it is very likely that you should go with that combination.

In the research carried out by Ohio State, RMS was determined to have the best correlation with the results that mixers would select by ear. This might be because RMS favors the low-end slightly. Or it might be because we always think the louder of 2 options is better. In either case, RMS is an objectively good target for the computer to use as a goal.

Because the optimizer works by switching the channel's existing polarity switches, it is always visible to the user; and the user can experiment with other settings if they like.

Best,
-Ben at Harrison