Comparison of frequency-fusion mechanisms for Binaural direction-of-arrival estimation for multiple speakers