Suppr超能文献

噪声环境下的语音识别、正常听力和听力受损情况下的时间与频谱分辨率

Speech recognition in noise, temporal and spectral resolution in normal and impaired hearing.

作者信息

Arlinger S, Dryselius H

机构信息

Department of Audiology, University Hospital, Linköping, Sweden.

出版信息

Acta Otolaryngol Suppl. 1990;469:30-7.

PMID:2356736
Abstract

A battery of tests for the evaluation of temporal (forward masking) and spectral (detection of frequency change; critical band for masking; psychoacoustic tuning curve) resolution in hearing with regard to the importance for speech recognition in noise has been applied to two groups of subjects. One group of 10 listeners had normal hearing, while the other consisted of 11 subjects with moderate, sloping sensorineural hearing loss. Significant differences were found between the two groups in all parameters except critical bandwidth for masking. Correlation analysis on the test results from the combined groups of subjects showed that speech recognition in noise was best correlated with the average hearing threshold levels at 2 and 4 kHz, followed by forward masking and psychoacoustic tuning curves at 4 kHz. The correlation coefficients were approximately 0.7, 0.6 and 0.55, respectively. Detection of frequency change and critical bandwidth show the lowest correlations to speech recognition in noise.

摘要

针对听力中时间分辨力(前掩蔽)和频谱分辨力(频率变化检测;掩蔽临界带宽;心理声学调谐曲线)对噪声中语音识别的重要性,已对两组受试者进行了一系列测试。一组10名听力正常的听众,另一组由11名中度、斜坡型感音神经性听力损失的受试者组成。除掩蔽临界带宽外,两组在所有参数上均存在显著差异。对两组受试者的测试结果进行相关性分析表明,噪声中的语音识别与2 kHz和4 kHz处的平均听力阈值水平相关性最好,其次是4 kHz处的前掩蔽和心理声学调谐曲线。相关系数分别约为0.7、0.6和0.55。频率变化检测和临界带宽与噪声中语音识别的相关性最低。

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验