SIMULATIONS

To show that the proposed method can extract the desired signal f₁(t) from the mixed signal f(t), we carried out three simulations using the following signals: (1) noisy AM harmonicity tone; (2) noisy synthesized vowel; and (3) noisy real speech. In simulation 1, f₁(t) was an AM complex tone, where F₀(t)=200 Hz, the tone's instantaneous amplitude was sinusoidal (10 Hz). In simulation 2, f₁(t) was a vowel /a/ synthesized by the log magnitude approximation (LMA) [4], where averaged $\overline{F_0(t)}=125$ Hz, and jitter was 5 Hz (from 123 to 128 Hz). In simulation 3, f₁(t) was a male vowel /a/ in the ATR database [7]. In all three simulations, f₂(t) was bandpassed pink noise. Five types of f(t) were used as simulation stimuli, where the SNRs of f(t) ranged from 0 to 20 dB in 5-dB steps.