| 摘要 |
A low complexity echo compensation is provided to reduce largely the load of process by executing the adaptation of filter coefficients of an echo compensation filtering unit using an optimal down sampling rate. A first analyzing filter bank(12) converts a part of a microphone signal into a microphone sub-band signal, which is down-sampled by a first down-sampling rate. A second analyzing filter bank(15) converts a loudspeaker signal into a loudspeaker sub-band signal, which is down-sampled by a second down-sampling rate lower than the first down-sampling rate so as to obtain a first loudspeaker sub-band signal. A memory stores the first loudspeaker sub-band signal. An echo compensation filtering unit(17) folds the first loudspeaker sub-band signal with an estimation value on the impulse response of a loudspeaker-room-microphone system, down-samples the folded first loudspeaker sub-band signal by a third down-sampling rate, obtains an echo signal contribution, which is down-sampled by the first down-sampling rate, echo-compensates for a microphone sub-band signal based on the echo signal contribution, and obtains the compensated microphone sub-band signal.
|
