Room Response Equalizer Based on Simplified RLS Adaptive Filter

Abstract

An equalizer of a room acoustic frequency responses is presented in this paper. Such equalizers often require the usage of a large number of weights adaptive filter that means a large computational complexity of the filter and the equalizer. The considered equalizer uses the modified x-filtered Recursive Least Squares (RLS) adaptive filter. The computationally efficient fast RLS adaptive filters are often unstable if the number of their weights is large. That is why such filters are not suitable for the application in the room equalizers. The Matrix Inversion Lemma (MIL) RLS adaptive filters are stable but they are computationally complex in such applications. The equalizer presented in this paper uses a simplified MIL RLS adaptive filter. The correlation matrix of the filter is diagonalized. Besides, each of a square submatrix at the diagonal of the simplified correlation matrix are inverted using the computationally efficient version of the MIL which takes into consideration the symmetrical structure of these submatrices. The simulation proves the operability of the equalizer and demonstrates the efficiency of the simplified equalizer comparting to its full version. The demonstration includes different pictures like the frequency responses, power spectral densities and signal waveforms.