This is fascinating - on two levels. First, that maintenance needs can be predicted by engine sound. Second - more mundanely - that this can be automated (via ML, or whatever).
This is effectively a virtual hybrid of the "knowing where to make the mark" craftsman with cybernetic systems control.