ASA Based Unitary Input Model for Sequential Processing of Speech Separation
Isao Nakanishi, Motohiro Ichikawa, and Naoto Sasaoka
Graduate School of Engineering, Tottori University, Tottori, Japan
Abstract—Speech separation based on auditory scene analysis (ASA) has been widely studied. In this study a computational ASA model, in which a mixed signal is sequentially decomposed into frequency signals using a modified discrete Fourier transform (MDFT), has been proposed. Four feature types of ASA are extracted from the decomposed frequency signals based on simple rules, and the decomposed frequency signals are regrouped by examining the characteristics of the extracted features. Finally separated speeches are obtained by adding the regrouped frequency signals in a modified inverse DFT. The separation performance of the proposed model is examined via computer simulations and subjective evaluations.
Index Terms—speech separation, auditory scene analysis, unitary input, sequential processing, modified discrete Fourier transform, subjective evaluation
Cite: Isao Nakanishi, Motohiro Ichikawa, and Naoto Sasaoka, "ASA Based Unitary Input Model for Sequential Processing of Speech Separation," International Journal of Signal Processing Systems, Vol. 7, No. 3, pp. 78-84, September 2019. doi: 10.18178/ijsps.7.3.78-84
Index Terms—speech separation, auditory scene analysis, unitary input, sequential processing, modified discrete Fourier transform, subjective evaluation
Cite: Isao Nakanishi, Motohiro Ichikawa, and Naoto Sasaoka, "ASA Based Unitary Input Model for Sequential Processing of Speech Separation," International Journal of Signal Processing Systems, Vol. 7, No. 3, pp. 78-84, September 2019. doi: 10.18178/ijsps.7.3.78-84