Modified Gain Toggling Approach to Controlling Acoustic Howling

The computational complexity requirement of an acoustic howling controller is a key factor determining its implementability, especially on resource-constrained hardware platform. This paper explores the development of a howling control algorithm that has very low computational cost and is implementable without requiring the use of hardware multipliers. The algorithm was analysed, modelled in a MATLAB® environment, and tested with speech and audio signals to obtain its performance indices. The modified gain toggling technique has a MSG Increase figure of 9.215 dB and 6.826 dB for speech and audio signals at a normalized switching frequencies of unity and 0.5 respectively. The average value of the sound quality index for both speech and audio signals are 1.6591 and 1.2118 respectively, and the asymptotic computational cost is O(1). Thus, it can be concluded that the developed algorithm has a reasonable MSG Increase figure and low computational cost at the expense of low sound quality indices for both speech and audio signals.