Technological influence on society has made man to depend more on human computer interactions in his day to day life. Removal of unwanted and rebound of sound from speech signal is must to enhance the clarity of the v...
详细信息
ISBN:
(纸本)9781509023998
Technological influence on society has made man to depend more on human computer interactions in his day to day life. Removal of unwanted and rebound of sound from speech signal is must to enhance the clarity of the voice in such interactions. A novel frequency domain adaptive algorithm to eliminate rebound sound effect in speech signal inside an auditorium is presented here. The forgetting Factor (lambda) on which steadiness of adaptive algorithm depends is kept constant in the suggested RLS algorithm. To enhance the computational power and confluence rate, a frequency domain dynamic filter algorithm is often utilized in abolishment of acoustic feedback. However, the correct step size is always an adjustment between the confluence rate and the stable-state performance. To solve this problem, Nitsch adapted a varying step-size procedure, where the step size is modified by spectral confluence state. A novel methodology to estimate the optimal step size is the subject of this paper. In addition, an artificial delay is inserted in the way of signal to estimate the time-domain system mismatch. Both flat stable-state misadjustment and quick confluence are achieved by the designed algorithm. Simulation results exhibit the effectiveness and the sturdiness of the designed method. At last, MATLAB is used to execute suggested algorithm and the experimentation outcomes demonstrated that the suggested average RLS and optimal step size is compared.
暂无评论