Automatic list for managing intros in radio broadcast

[Mastervideo]

Being a broadcasting enthusiast, I noticed that broadcasters often have to manually search for the exact point where the artist starts singing. When it comes to doing this on a large number of songs, it becomes a really long job and so I thought: could we not use UVR? The system should have a model that allows to scan a certain folder, analyze the vocal part separately and provide as output a txt or csv file in which the time in seconds between the start of the single file and the sung part is indicated. I am not a programmer, but I think it is not a bad idea. So, with the csv file it becomes easier to import these cues into playout programs. Is it feasible? Thanks.