Deep Learning Based Personalized HRTF Estimation Using Anthropometric parameters and Ear Images

Abstract

This paper proposes a personalized head-related transfer function (HRTF) estimation method based on deep neural networks by using anthropometric measurements and ear images. A feedforward deep neural network (DNN) using the anthropometric measurements regarding the head and torso and a convolutional neural network (CNN) using the ear images are constructed. After that, the outputs of these two networks are merged into another DNN for estimation of the personalized HRTF. To evaluate the performance of the proposed method, objective and subjective evaluations are conducted. For the objective evaluation, the root mean square error (RMSE) and the log spectral distance (LSD) between the reference HRTF and the estimated one are measured. Consequently, the proposed method provides the lower RMSE and LSD than the DNN-based method using anthropometric data without pinna measurements, respectively. Next, a sound localization test is performed for the subjective evaluation. As a result, it is shown that the proposed method can localize sound sources with higher accuracy than the average HRTF method and DNN-based method, respectively. In addition, the reductions of the front/back confusion rate are achieved by the proposed method, compared to the average HRTF method and DNN-based method, respectively.