| 摘要 |
A method is provided of extracting desired signals s<SB>t</SB> from contaminated signals y<SB>t</SB> measured via respective communication channels. The system comprising the desired signals s<SB>t</SB> and the channels is modelled as a state space model. In the model, the desired signals have time-varying characteristics which vary more quickly than second time-varying characteristics of the channels. The method may be employed to extract individual speech signals from speakers s<SB>t</SB><SP>(1)</SP> and s<SB>t</SB><SP>(2)</SP> in a room 1, contaminated signals y<SB>t</SB> being sampled by microphones 6, 7, 8 and processed by a computer 9 which extracts the individual speech signals for supply to respective loud speakers 13, 14.
|
