In feedback active noise control, optimization algorithms，such as H2/H∞  robust control，are difficult to find the optimal solution in practical design because of its complex parameter and initial value selection. Adaptive feedback active noise control system, based on internal model control structure, is adaptive to different circumstances but it is necessary to consider the system stability and control noise amplification effectively. In order to reduce the time delay, the digital feedback control generally uses a higher sampling rate, which leads to a wider frequency band and useless noise reduction at high frequency. In this paper, a new algorithm using a weighting filter to adjust the noise reduction of different frequency bands is proposed. A penalty term according to different weighting filter is introduced into the iteration of the traditional FxLMS active noise control. Compared with the previous waterbed effect tuning algorithm, double-gradient algorithm, the new algorithm can achieve a higher noise reduction at low frequency while tuning the waterbed effect and achieve a good balance between the low-frequency noise reduction and waterbed suppression. In addition to tuning waterbed effect, the magnitude in high frequency of the weighting filter is added to restrict the noise reduction at high frequency to ensure the larger noise reduction at low frequency. The application of the algorithm in high frequency constraint enables the feedback active noise control system to achieve a larger noise reduction at low frequency while tuning the waterbed effect in a wide frequency band.