In Jutten’s blind separation algorithm, symmetrical distribution and statistical independence of the signal sources are assumed. When they are not satisfied, the learning proce...
— This paper presents an eigenvector algorithm (EVA) derived from a criterion using reference signals, in which the EVA is applied to the blind source separation (BSS) of instant...
This paper examines the performance of several source separation systems on a speech separation task for which human intelligibility has previously been measured. For anechoic mixt...
Michael I. Mandel, S. Bressler, Barbara G. Shinn-C...
Boosting algorithms build highly accurate prediction mechanisms from a collection of lowaccuracy predictors. To do so, they employ the notion of weak-learnability. The starting po...
— We present a frequency-domain method of noisy convolutive source separation, where we extends the SEONS algorithm [2] that jointly exploits the nonstationarity and temporal str...
Inseon Jang, Kyeongok Kang, Sangki Kim, Seungjin C...