Currently no. If you listen to them they sound terrible. But I made a GitHub issue for the ability to change voices, and I hope to be shipping that soon. Autodetection is a bit harder but probably not by much.