Abstract:

This study examined the effects of noise type, signal-to-noise ratio (SNR), and masking mechanism on speech perception by comparing keyword recognition and sentence comprehension across different noise environments. Results showed a significant interaction between noise type and task performance. Babble noise exacerbated the difference between tasks via semantic informational masking, exerting a greater disruption on sentence comprehension tasks that rely heavily on semantic processing. In modulated speech-shaped noise, both tasks exhibited similar dynamic masking release by utilizing “temporal dips”. In contrast, long-term speech-shaped noise primarily functioned through energetic masking, yielding task differences only at extreme SNRs. The weak correlation between the two tasks suggests they play a complementary role in clinical assessment. The study provides theoretical insights for the assessment of speech recognition capabilities in complex noise environments.

Key words: Mandarin speech perception, noise, informational masking, masking release from temporal modulation

摘要：

本研究对比了不同噪声环境下关键词识别任务与语句理解任务的表现，系统考察了噪声类型、信噪比及掩蔽机制对言语感知的影响。结果显示，噪声类型与测试任务存在显著交互作用：人声噪声通过语义信息掩蔽显著扩大了任务间差异，对语义需求高的语句理解任务干扰更大。在语谱动态噪声环境下，两种任务均能从“时域间隙”中获得相似的动态掩蔽释放效应；而语谱稳态噪声主要产生能量掩蔽，仅在极端信噪比下引发任务间差异。不同噪声条件下两类任务表现出的弱相关性，支持其在临床评估中的互补作用。本研究为复杂噪声环境中言语识别能力的评估提供了理论借鉴。

关键词: 普通话言语感知, 噪声, 信息掩蔽, 动态掩蔽释放