一种基于最小二乘法的鲁棒性耳机均衡技术

Robust Headphone Equalization Technique Based on Least Squares Method

在基于耳机的双耳声重放中,需要对耳机到鼓膜的传递函数(HpTF)进行均衡,可使聆听者在耳机重放时获得自然的听觉感受。但在实际应用中,均衡效果对耳机测量位置变化非常敏感,系统鲁棒性差。本文提出一种基于最小二乘法的鲁棒性耳机均衡技术,该方法通过提取HpTF的频谱特征来自动地估计正则化参数的取值,在实现鲁棒性控制的同时,可达到理想的均衡效果。本文对提出的方法进行了数值仿真,并通过与传统最小二乘法、complex smoothing方法进行比较,验证了其有效性。

In binaural sound reproduction over headphones, the headphone-to-eardrum transfer functions(HpTFs) need to be equalized such that listeners can obtain a natural perceptual experience in headphone reproduction. However, in practice, the effects of equalization are sensitive to the variations of the headphone measurement positions, and the system suffers from poor robustness. In this paper, a robust headphone equalization technique based on the least squares method is proposed, which automatically estimates the value of the regularization parameter by extracting the frequency spectrum characteristic of HpTF, and obtains the ideal equalization effects while achieving robust control. This paper presents the numerical simulations of the proposed method compared with the conventional least squares method and complex smoothing method, validating the effectiveness of the proposed method.