Experimental stimuli

For experimental stimuli, we will assume f₁(t) is a sinusoidal signal and f₂(t) are two types of noise where f₂₁(t) is an AM bandpassed noise and f₂₂(t) is a bandpassed random noise, as follows:

 

$$F_{\rm {BP}}(g_1),$$ f₁(t) = (26)

$$\left\{ \begin{array}{l} 1200\sin(2\pi f_0 t), \\ \qquad 0.3+T_m \leq t \leq 0.7+T_m, \\ 0, \quad \mbox{otherwise} \\ \end{array}\right.$$ g₁(t) = (27)

$$\sum_{f=f_0-500}^{f_0+500}E_M(t)\sin(2\pi ft+R(f)),$$ f₂₁(t) =

$$\qquad \qquad 0 \leq t \leq 1.0,$$ (28)

$$\sum_{f=f_0-500}^{f_0+500}E_R(f,t)\sin(2\pi ft+R(f)),$$ f₂₂(t) =

$$\qquad \qquad 0 \leq t \leq 1.0,$$ (29)

where $F_{\rm {BP}}(\cdot)$ is a bandpass filter with a center frequency f₀ and a bandwidth of 23 Hz (as shown in Fig. ), f₀=600 Hz, $T_m=0.0125m, m=0,1,\cdots, 9$, and R(f) is an uniform random within $[-\pi,\pi]$. Here, E_M(t) and E_R(f,t) are amplitudes in which white noise lowpass filtered at 30 Hz and added to the bias value to prevent over modulation. Note that the amplitude E_R(f,t) fluctuates independently in frequency region f, and that the power of noise is adjusted so that $\sqrt{f_{21}(t)^2/f_{22}(t)^2}=1$. Here, the bandwidth of noise is 1 kHz and the SNR between a sinusoidal signal and the bandpassed noise is -8.5 dB. These mixed signals are shown in Fig. .

The two types of mixed signals are f_M(t)=f₁(t)+f₂₁(t) when f₂(t)=f₂₁(t) and f_R(t)=f₁(t)+f₂₂(t) when f₂(t)=f₂₂(t). When a human hears these mixed signals, he can hear the sinusoidal signal from f_M(t) caused by CMR; however cannot hear the sinusoidal signal from f_R(t) caused by Masking. These mixed signals are shown in Fig. .

In this simulation, segregation is done using 10 mixed signals which are made by varying the onset of f₁(t), $m=0, 1, \cdots, 9$ in Eq. ().