A joint sparsity and linear regression based method for customization of median plane HRIR

In this paper, a joint sparsity and multiple linear regression(MLR) based method for computing customized median plane Head Related Impulse Response(HRIR) is proposed. A novel sparsity and MLR based method are first used to identify the dominant pinna parameters individually. Subsequently, parameters obtained by both these individual methods are combined to obtain smaller set of pinna parameters using the principle of majority voting. The HRIRs for an unknown subject are then customized using this meaningful subset of pinna parameters. Experiment on pinna spectral notch extraction and RMSE analysis are used to illustrate the improved performance provided by the proposed method.