Personalized HRTF prediction based on LightGBM using anthropometric data

Abstract

This paper proposes a personalized head-related transfer function (HRTF) prediction method based on LightGBM using anthropometric data. Considering the overfitting problems of the current training-based prediction methods, we use LightGBM and a specific network structure to prevent over-fitting and enhance the prediction performance. By decomposing and combining the data to be predicted, we set up 90 LightGBM models to separately predict the 90 instants of HRTF in log domain. At the same time, the method of 10-fold cross-validation is used to score the accuracy of the model. For models with scores below 80 points, Bayesian optimization is used to adjust model hyperparameters to obtain a better model structure. The results obtained by LightGBM are evaluated with spectral distortion (SD) which can show the fitting error between the prediction and the original data. The mean SD values of both ears on the whole test set are 2.32 dB and 2.28 dB respectively. Compared with the non-linear regression method and the latest method, SD value of LightGBM-based method relatively decreases by 83.8% and 48.5%.