噪声频谱特性对正常听力成人词汇识别影响

目的　探讨噪声频谱特性对词汇识别的影响作用。方法　31名听力正常成人参与研究，研究采用汉语普通话词汇相邻性测试，分别获得言语谱噪声（speech spectrum-shaped noise，SSN）和多人谈话噪声（babble noise，BN）下不同信噪比（signal noise ratio，SNR）的言语识别率，并计算词汇识别率为20%、50%和80%时所需要SNR和斜率，进而分析噪声频谱特性对词汇识别的影响作用。结果　SSN条件下单音节词和双音节词的言语识别阈，即正确识别率为50%所对应的SNR（SNR50）分别为-4.83 dB和-7.63 dB（t 50=11.918，P＜0.05），对应斜率分别为5.61和11.14（t 50=-7.006，P＜0.05)。BN条件下单音节词和双音节词的SNR50分别为-3.34 dB和-5.52 dB（t 50=8.860，P＜0.05)，对应斜率分别为6.04和9.28（t 50=-4.316，P＜0.05）。统计显示SSN和BN下单、双音节词识别阈均有统计学差异（t 50M=-5.037，t 50D=-10.275，P＜0.05），两类噪声下双音节词的斜率有统计学差异（t 50D=2.920，P＜0.05）。结论　BN噪声频谱掩蔽效应比SSN更强。在两种噪声下，单音节词均需要更高的SNR，才能达到和双音节词相同的词汇识别。

Effects of noise spectrumon word perception in normal hearing adults

OBJECTIVE The purpose of the present study was to investigate the effects of noise spectrum, including speech spectrum-shaped noise(SSN), and 4-talker babble noise(BN) on spoken word perception in noise in normal hearing adults. METHODS In all, thirty-one native Mandarin speaking normal hearing adults (age ranged from n 20 to 26 years) participated in the study.The evaluation material was M-LNT, each participant was tested in SSN and BN, respectively. P-SNR was obtainedfor each participant. RESULTS The SNR50 of disyllable and monosyllable in SSN were -7.63 dB and -4.83 dB, and the corresponding rate ratio were 11.14 and 5.61, respectively. The SNR50 of disyllable and monosyllable in BN were -5.52 dB and -3.34 dB, the corresponding rate ratio were 9.28 and 6.04, respectively. There were significant differences of SNR50 between disyllable and monosyllable in SSN and BN(P<0.05). CONCLUSION Compared with SSN, greater masking effect was observed in BN in normal hearing adults.Specifically, the P-SNR function of M-LNT in BN shifted to the rightand increased slower than SSN. Similar lexical effect was found in each typeof noises. That is, compared to disyllabic word, better SNR was needed for monosyllabic word to obtained same perception correct.