几类典型环境声的主观评价及感知特性分析

近年来,通过“注入”调控声以降低交通噪声烦恼感的声频注入法受到广泛关注。以交通噪声调控研究为背景,通过成对比较评价了4类典型声音(实验一)和4类典型交通噪声(实验二)的烦恼感。结果表明,有调声(纯音和复音)烦恼度最高,自然声最低(海潮声最佳),蓝色噪声是仅次于海潮声令人感觉舒适的声音;被试对交通噪声和白噪声的评价存在明显的分类偏好。分析心理声学特征发现人对声音的感知依赖于多方面因素,但声刺激的某一因素(如粗糙度或音调特别高)特别突出则会引起极大的反感。构建不相似度二维感知空间,维度1反映了声音类型间的差异,维度2表征了被试对不同类型声音的烦恼度评价;并且通过相关分析发现它们与谱结构参量相关性较强。接下来的研究中,可以通过调整交通噪声的谱下降值和时域上升时间等参量使其谱结构更接近于自然声,从而降低噪声烦恼度。

Subjective evaluation and perceptual characteristics analysis of several typical ambient sounds

Traffic noise is the main source of noise pollution. In recent years, the audio injection method to reduce traffic noise annoyance by "injecting" control sound has received widespread attention. Based on the research of traffic noise control, this paper evaluates the annoyance of 4 kinds of ambient sound (experiment 1) and 4 kinds of traffic noise (experiment 2) through paired comparison. The results show that the tone (pure tone and FM tone) has the highest annoyance among the selected sounds, natural sound is the most comfortable (the tide has the lowest annoyance), and the blue noise is only inferior to the sea tide sound. The subjects have obvious classification preference of traffic noise and white noise. The analysis of psychoacoustic parameters found that the human''s perception of sound is related to various factors. However, if a certain factor (such as extremely rough or high pitched) is particularly prominent in the sound stimulus, it will cause a great deal of annoy. A two-dimensional perception space of environmental sound dissimilarity is constructed. The first dimension reflects the differences between the types of stimuli, which is mainly related to the spectral roll-off of the sound in experiment 1 and the specific loudness mean value (the first Bark) in experiment2. The second dimension represents the annoyance of subjects on different types of sounds. For experiment 1, the temporal parameter attack estimation has a strong correlation with this dimension, and the corresponding parameter of experiment 2 is the 1/3 octave band sound level with the center frequency of 200Hz. In the following research, the spectral structure of traffic noise can be closer to natural sound by modifying the e spectral roll-off and attack estimation, thus reducing noise annoyance.