Revisiting the security of speaker verification systems against imposture using synthetic speech