Blind separation of speech mixtures via time-frequency masking
IEEE Transactions on Signal Processing
Survey of clustering algorithms
IEEE Transactions on Neural Networks
| Hi-index | 0.00 |
We recently proposed several time-frequency blind source separation methods applicable to attenuated and delayed mixtures. They were limited by heuristics and only processed integer-sample time shifts. In this paper, we propose and test improvements of these approaches based on modified clustering techniques for scale coefficients and a new iterative method to estimate possibly non-integer and large delays.