Exploring the relationship between voice similarity estimates by listeners and by an automatic speaker recognition system incorporating phonetic features.