噪声信号声品质评价及分析方法研究
文献类型：学位论文
| 作者 | 焦风雷 |
| 学位类别 | 博士 |
| 答辩日期 | 2005 |
| 授予单位 | 中国科学院声学研究所 |
| 授予地点 | 中国科学院声学研究所 |
| 关键词 | 声品质 车内噪声 主观评价 客观评价 心理声学 非度量多维尺度分析 |
| 中文摘要 | 噪声信号的声品质研究是环境噪声一个新的研究领域，其研究成果在环境噪声、产品噪声等多项领域有广泛的应用前景。该研究在国际上仅有十多年的历史，目前还不成熟。国内这方面的研究正处于起步阶段。本论文以匀速行驶条件下等响处理后的车内噪声为研究对象，采用成对比较法、参考语义细分法以及非度量多维尺度分析（NMDS）系统进行噪声声品质的主客观评价及其分析方法研究。成功运用NMDS方法，将评价者先进行分类处理，进而揭示出一致性较强的车内噪声声品质评价中的潜在结构信息，并简化了模型分析方法。提出了采用评价者之间相关性的NMDS作为主观评价实验数据检验，可减少实验量、实现整体数据的检验。提出了采用多元线性回归方法在小样本条件下从众多客观参量中提取冗余度小、解释量高的特征量。揭示出可懂度、尖锐度、平均特征响度刘·于声品质评价的重要作用。基于巧个样本的成对比较法，采用人群分类处理建立了偏好性模型，对该模型用40个样本的参考语义细分法进行校验。该模型相比国际上噪声声品质的偏好性模型拟合优度较差的情况，有了较大提高。最后提出了优化的噪声声品质主观及客观评价研究方案，对进一步的环境噪声声品质主观客观评价研究具有明确的指导意义。 |
| 英文摘要 | Assessment of Sound quality is a very newly field in environmental acoustics since the end of last century, and finds its growing and extensive application in automobile, household electric appliances etc. Currently research of sound quality is an international academic hot spot, while domestic research in this field is still at the beginning stage. Starting from sample of loudness-adjusted car interior noise at constant driving condition, with pair-wise comparison, semantic differential method and nonmetric multidimensional scaling (NMDS), research on subjective evaluation and analysis methods for sound quality is carried out. It is presented that NMDS is an effective tool to explore subjective assessment of sound quality. Based on the key process, classifying by correlation between test people, subjective perceptive structure and psychophysical foundation of stronger consistent car interior noise and can be opened out. And it will lead to a succinct preference model of noise quality. Moreover, it is proposed that NMDS of correlation between test people can be taken as data validation for subjective assessment experiment with fewer work quantity of subjective evaluation and more effectual validation for total data. In addition, it is proposed advanced statistical data processing for both grouped pair-wise comparison and semantic differential method with more reasonable scientificity. A systematical research on objective evaluation and analysis methods for sound quality are firstly performed in China. And the effect of characters of physicoacoustics, psychoacoustics and statistical on assessment of sound quality are investigated. Base on analysis between several multivariate statistical methods, it is put forward that multivariate linear regression can be used to extract less information redundancy and larger interpretive feature from too many parameters at a few samples condition. It's found that speech intelligibility index, sharpness, and average special loudness exert major effect on assessment of car interior noise quality. Based on pair-wise comparison of 15 samples, preference mode is established. And the mode is validated by data from semantic differential method of 40 samples. Compared with general preference modes in environmental acoustics, the preference mode proposed by this disquisition has larger goodness of fit. According to results of two aspects, an optimizing principle of subjective and objective assessment of sound quality research is suggested. This constructive suggestion is important to improve research on assessment of sound quality in environmental acoustics. |
| 语种 | 中文 |
| 公开日期 | 2011-05-07 |
| 页码 | 119 |
| 源URL | [http://159.226.59.140/handle/311008/908]  |
| 专题 | 声学研究所_声学所博硕士学位论文_1981-2009博硕士学位论文 |
| 推荐引用方式 GB/T 7714 | 焦风雷. 噪声信号声品质评价及分析方法研究[D]. 中国科学院声学研究所. 中国科学院声学研究所. 2005. |
入库方式： OAI收割
来源：声学研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。