Conclusions

This paper proposed a new method of extracting the desired speech from noisy speech precisely even in waveforms by using constraints on the continuity of instantaneous phases as well as those on the continuity of instantaneous amplitudes and fundamental frequencies.

In order to show that the proposed model can extract real speech from noisy speech precisely even in waveforms, we demonstrated one evaluation and three simulations of segregating two acoustic sources. The result of evaluation showed that all constraints related to the four regularities are useful in order to segregate the desired vowel from a noisy vowel. The results of the three simulations showed that the proposed method can segregate the desired vowel from a noisy vowel precisely even in waveforms. It was also shown that the proposed method can precisely segregate the desired signal from the simultaneous signals using the instantaneous amplitude and phase.

Future work includes (1) reconsidering the order of the polynomial approximation of C_k,R(t) and D_k,R(t) for vowel segregation and (2) improving the proposed model so that it can be applied to consonants-vowel segregation.