Date: | Friday, Sep 24, 2004 |
---|---|
Time: | 13:30 - 14:50 |
Place: | L1 Lecture room |
Language: | English (oral presentation), English/Japanese (question) |
Chairperson: | Tsuyoshi Iwagaki (Computer Design and Test Lab.: D3),
Yuichiro Kanzaki (Software Engineering Lab.: D2) |
In this presentation, in order to improve the decomposition performance, we propose the learning algorithm combining frequency-domain ICA with projection back processing and SIMO-ICA. To evaluate its effectiveness, decomposition experiments are carried out under a reverberant condition for the different directions of sources. The experimental results reveal that the decomposition performance of the proposed method is superior to that of the conventional method.