How can I put different weight on two channels in a keras CNN Autoencoder model for timeseries?

Question

Is is possible to put more "attention"/"weight"/"importance" on one channel and less on the other channel in a machine learning model in keras? I.e