One main obstacle in speech recognition is what said "robustness". This paper focus on one popular idea in antagonizing speech system vulnerability-channel normalization, and presents a new normalization algorithm- Multi-Layer Channel Normalization (MLCN), which exploits the recursive compensation progress in two domains- spectral domain and cepstral domain- to depress different noises, so that the more robust speech representation is achieved. Experimental results of our gallina system demonstrate the validity of our new algorithm.