Considering that setting cues, intro's and Hooks is a very time consuming process. it would be a great feature to have AI analyzing tracks and setting these points in the track database. since AI is able to seperate vocals from music, this should be possible.
Comment on this feature request:
Jason Allen commented
While this is technically possible, it is unreliable - certain instruments can get detected as voices. In addition, it is very computationally expensive, taking up to 30 seconds per song on a reasonably performant computer. With large libraries this would take a very long time.
Maarten Schutjes commented
1. You can save a lot of time by just scanning the first minute. Ifvocal start not found refer to manual
2 You can do it with software like LALALA API.
3 It is insanely expensive to use.. so don't do it ; )