Sprenger Kristin, Brand Thomas
Medizinische Physik, Carl von Ossietzky Universität Oldenburg, Oldenburg 26111, Germany.
Cluster of Excellence "Hearing4all," Carl von Ossietzky Universität Oldenburg, Oldenburg 26111, Germany.
J Acoust Soc Am. 2025 Feb 1;157(2):1126-1140. doi: 10.1121/10.0035801.
This study evaluated three models quantifying context processing in human speech recognition. We presented everyday sentences with high semantic predictability and closed word set (matrix) sentences with low semantic predictability to listeners with normal hearing and listeners with hearing loss. Sentences were presented in quiet, in steady-state noise, and in speech-like modulated noise. Reverberated and clean speech were used. We evaluated whether the models are suited for testing the following hypotheses: (1) "Listeners with hearing loss compensate reduced auditory input by relying more strongly on speech context." (2) "Speech context is used to compensate missing information differently in modulated noise than in steady state noise or in quiet, with a smaller difference in context effects for listeners with hearing loss." (3) "Context processing explains why matrix sentences are much more robust against reverberation than everyday sentences." Significance testing was performed using bootstrapping. Only one of the context models showed significant differences in context processing between normal-hearing and hearing-impaired listeners for the different sentence types and for different listening conditions. No differences in context processing were found between stationary and modulated noise. To explain the behavior in reverberation, it is important to distinguish between a priori context and a posteriori context.
本研究评估了三种量化人类语音识别中语境处理的模型。我们向听力正常的听众和听力损失的听众呈现了语义可预测性高的日常句子以及语义可预测性低的封闭词集（矩阵）句子。句子分别在安静环境、稳态噪声和类似语音的调制噪声中呈现。使用了有混响的语音和纯净语音。我们评估这些模型是否适合检验以下假设：（1）“听力损失的听众通过更强烈地依赖语音语境来补偿减少的听觉输入。”（2）“与稳态噪声或安静环境相比，语音语境在调制噪声中用于补偿缺失信息的方式不同，听力损失的听众的语境效应差异较小。”（3）“语境处理解释了为什么矩阵句子比日常句子对混响的抵抗力要强得多。”使用自抽样法进行显著性检验。对于不同的句子类型和不同的聆听条件，只有一种语境模型在听力正常和听力受损的听众之间的语境处理上显示出显著差异。在稳态噪声和调制噪声之间未发现语境处理的差异。为了解释在混响中的行为，区分先验语境和后验语境很重要。
