5th IEEE Hiroshima Student Symposium, December 2003, Hiroshima, Japan A Method for Segregation of Speech Signals E. MOLINE (Undergraduate student in ENSIETA, Brest - FRANCE) J. TSUTSUI (Master Student, Dept. of Electronic and Control Systems Eng., Shimane University, JAPAN) T. OKAMOTO (Master Student, Dept. of Electronic and Control Systems Eng., Shimsne University, JAPAN) Ali MANSOUR ( Lab. E3I2, ENSIETA, Brest, FRANCE) M. KAWAMOTO (Dept. of Electronic and Control Systems Eng., Shimane University, Shimane, JAPAN) Y. House (Dept. of Electronic and Control Systems Eng., Shimane University, Shimane, JAPAN) ABSTRACT This paper deals with the problem of blind separation of under-determined or over-completes mixtures (i.e. more sources than sensor) and more particularly in the special case of two speech signals and one sensor. After a introduction to the problem of blind separation of sources, an algorithm based on time-frequency representations (TFR) is presented. Finally, some experiments are conducted and some experimental results are given.