Mixture Models Applied to the Estimation of Mixing Parameters in Multi-Channel Blind Source Separation Algorithms
César Clares-Crespo, Roberto Gil-Pita, Manuel Rosa-Zurera, Joaquín García-Gómez, and
Inma Mohíno-Herranz
University of Alcalá, Department of Signal Theory and Communications, 28805-Alcalá de Henares, Madrid, Spain
Abstract—The Sound Source Separation (SSS) problem is treated depending on the premises that characterize a specific separation problem. Within some that carry out a blind separation the limitation come from the acoustic scene with the reverberation. It is a need to look for solutions focusing on these limitations. For methods based on a time-frequency approach where an estimation of the parameters of the mixture is required, we study different ways of estimation based on different probability density functions that can perform better in more disadvantaged acoustic scenes.
Index Terms—audio signal processing, sound source separation, microphone array, mixture model, speech enhancement
Cite: César Clares-Crespo, Roberto Gil-Pita, Manuel Rosa-Zurera, Joaquín García-Gómez, and Inma Mohíno-Herranz , "Mixture Models Applied to the Estimation of Mixing Parameters in Multi-Channel Blind Source Separation Algorithms," International Journal of Signal Processing Systems, Vol. 7, No. 3, pp. 85-91, September 2019. doi: 10.18178/ijsps.7.3.85-91
Index Terms—audio signal processing, sound source separation, microphone array, mixture model, speech enhancement
Cite: César Clares-Crespo, Roberto Gil-Pita, Manuel Rosa-Zurera, Joaquín García-Gómez, and Inma Mohíno-Herranz , "Mixture Models Applied to the Estimation of Mixing Parameters in Multi-Channel Blind Source Separation Algorithms," International Journal of Signal Processing Systems, Vol. 7, No. 3, pp. 85-91, September 2019. doi: 10.18178/ijsps.7.3.85-91