In the tele-conference, the speech recognition system is mainly affected by the reverberation and not affected by the noise. The model for receiving signals adopted by most time delay estimation methods are additive noise model, which are not suitable for
展开▼