Modified delay compensated adaptive gamma filter for noise cancellation in telephonic speech

Noise removal continuous to be a challenging domain of work despite of several techniques which are proposed to overcome this issue are found to be efficient and reliable. In mobile communications, the time varying nature of signals and noise makes this issue more challenging. Among conventional signal processing techniques, finite impulse response (FIR) and infinite impulse response (IIR) structures provide their own set of advantages and disadvantages. Though gamma filters in FIR and IIR forms have been suggested for noise cancellation in speech, it is bottlenecked due to issue such as longer convergence time, delay generation and inability to provide phase variation compensation. This paper presents certain techniques to overcome this issue. Firstly, a modified gamma filter with a feedback structure is used to improve MSE convergence and provide better phase variation correction. The second method uses a transfer function approach for phase correction of the filtered signal. Experiments show that the MSE and phase difference(in degree) between the filtered signal and the desired signal are lower for the proposed structure than the MSE and phase difference values obtained for the FIR filter and original gamma filter structure. Thus the suggested techniques have been found to be more reliable and efficient.