一种新的艺术嗓音客观评价方法

A new objective evaluation method of artistic voice

针对目前艺术嗓音评价效率低、主观性强的问题,提出了一种基于卷积神经网络的嗓音质量客观评价方法。在该方法中,将音频信号转化为一定尺寸的梅尔声谱图,并构建了一种多层CNN网络架构的图像特征模型,使得艺术嗓音客观评价问题转化为图像分类问题。实验表明,通过深度学习方法客观评价艺术嗓音质量,相比于已有提取声学参数和机器学习的评价方法,准确率有一定提高,为客观高效地评估艺术嗓音提供了一种新方法,具有较高的应用价值。

To solve the problems of low efficiency and strong subjectivity in the evaluation of artistic voice,this paper proposes an objective evaluation method of voice quality based on convolution neural network.In this method,the audio signal is firstly transformed into a certain size of Mel spectrogram,and then a feature model of image of multi-layer CNN network is established,thereby the objective evaluation problem of artistic voice is transformed into a problem of image classification.The experimental results show that the accuracy of the deep learning method is higher than that of the existing approaches such as the extraction of acoustic parameters and the machine learning.It provides a new method for the objective and effective evaluation of artistic voice,and has high application value.