流媒体音乐质量客观评测方法设计

Novel Objective Method for Evaluating the Quality of Streaming Audio

针对音频感知质量评价模型(PEAQ)在流媒体音乐质量评测中存在的问题,提出一种针对流媒体音乐质量客观评测的框架——流媒体音频感知质量评价模型(PESAQ).用PESAQ设计对齐算法消除流媒体音乐中的抖动,并使用PEAQ评测编解码、噪声等造成的损伤,同时引入基于梅尔倒谱系数的动态时间规整最小距离对抖动造成的损伤进行评测,最后综合两种损伤的评测结果得到流媒体音乐的客观质量评分.主客观对比实验表明,PE-SAQ对流媒体音乐质量的评测结果比PEAQ更接近主观实验结果.

An objective method for evaluating the quality of streaming audio called perceptual evaluation of streaming audio quality（PESAQ）was presented.PESAQ firstly removed the jitters in the streaming audio by an alignment algorithm.Secondly,the degradation between reference audio and the aligned streaming audio was measured by perceptual evaluation of audio quality（PEAQ）,and the result was called objective difference grade（ODG）.Thirdly,the degradation caused by jitters were measured by the dynamic time warping（DTW）minimum distance,and the result was called DTW-ODG.Finally,the quality of the streaming audio was derived by summing ODG and DTW-ODG.The experiment results prove that PESAQ provides closer results to the subjective results than PEAQ for streaming audio quality evaluation.