Nonindividual HRTF(1993)

• Goal:

      - use a good localizer's HRTF as a universal HRTF
      - need not to customize HRTF for every individual



• Result:

      - an increase in rate of front-back confusion
      - slightly degraded for good localizer using HRTF from
        another good localizer
      - large errors made for good localizer using HRTF from a
        poor localizer
      - poor anyway for a poor localizer

• Solving: Y1 = Y2

      - X2 = X1LF/H
      - define T = LF/H



• Goal:

      - measure T for every subject in 144 source positions
      - given x1(t), then X2 = X1T, and x2(t) is available



• Experiments:

      - measure Y1, X1, Y2, X2(可測量) 
      - LFM = Y1/X1, and HM = Y2/X2
      - T = LFM/(HM) = LF/M
      - Transfer function were measured from both ears at 144
        source positions
      - elevation -36, -18, 0, 18, 36, 54 degrees(6 divisions)
      - azimuth - 15 degree each step(24 divisions)  so 6 * 24 = 144 
      - See Fig. 1, 2, 3, 4 and 6.

	L: loudspeaker transfer function
 	F: free-field-to-eardrum transfer function
	H: headphone-to-eardrum transfer function

	X1(t): 揚聲器信號
	X2(t): 耳機信號
	Y1(t): 耳膜收到 X1(t) 之結果信號
	Y2(t): 耳膜收到 X2(t) 之結果信號