Alan Shaw is obsessed about accurately reproducing the human voice, and he brings up the topic regularly on Harbeth's forum. Shaw's reasoning is that our hearing developed from millions of years of evolution to be hyper sensitive at picking up subtle nuances in the human voice. Therefore speaker performance is especially critical in the human vocal range where our hearing is most discerning. He has talked about using recordings of spoken speech during their engineering development and how it highlights qualities that may not be apparent from music.
This focus explains the strengths and weaknesses of Harbeth speakers. If you can get the human voice right, a lot of things come along with it. It is the reason for the famed Harbeth midrange and why it is particularly good at reproducing unamplified acoustic instruments. However, they haven't given the same attention to reproducing electronic sounds and dynamics such as in rock music.