Noise radars possess several advantageous properties, including low probability of detection and identification, especially when working in continuous-wave mode. One of the main drawbacks of such radars is the occurrence of the masking effect, when a weak target echo is masked by the sidelobes of a strong target echo. The sidelobes of the ambiguity function emerge on the level of the time–bandwidth product below the main peak and are spread in the entire range–Doppler plane. While most approaches to mitigate the masking effect focus on the processing of the received signal, it is possible to move the computation burden to the waveform design phase. This paper describes a filter-based method of creating noise-like waveforms that have very low sidelobes in the area of certain range and Doppler shifts. Both theoretical analysis and measurement results are presented and compared with the lattice filter method of masking effect cancellation.