Ho Cheng-Yu, Li Pei-Chun, Chiang Yuan-Chuan, Young Shuenn-Tsong, Chu Woei-Chyn
Department of Biomedical Engineering, School of Biomedical Science and Engineering, National Yang-Ming University, Taipei City, Taiwan, R.O.C; Holistic Education Center, Mackay Medical College, New Taipei City, Taiwan, R.O.C.
Department of Audiology and Speech-Language Pathology, Mackay Medical College, New Taipei City, Taiwan, R.O.C.
PLoS One. 2015 Apr 2;10(3):e0120977. doi: 10.1371/journal.pone.0120977. eCollection 2015.
Binaural hearing involves using information relating to the differences between the signals that arrive at the two ears, and it can make it easier to detect and recognize signals in a noisy environment. This phenomenon of binaural hearing is quantified in laboratory studies as the binaural masking-level difference (BMLD). Mandarin is one of the most commonly used languages, but there are no publication values of BMLD or BILD based on Mandarin tones. Therefore, this study investigated the BMLD and BILD of Mandarin tones. The BMLDs of Mandarin tone detection were measured based on the detection threshold differences for the four tones of the voiced vowels /i/ (i.e., /i1/, /i2/, /i3/, and /i4/) and /u/ (i.e., /u1/, /u2/, /u3/, and /u4/) in the presence of speech-spectrum noise when presented interaurally in phase (S0N0) and interaurally in antiphase (SπN0). The BILDs of Mandarin tone recognition in speech-spectrum noise were determined as the differences in the target-to-masker ratio (TMR) required for 50% correct tone recognitions between the S0N0 and SπN0 conditions. The detection thresholds for the four tones of /i/ and /u/ differed significantly (p<0.001) between the S0N0 and SπN0 conditions. The average detection thresholds of Mandarin tones were all lower in the SπN0 condition than in the S0N0 condition, and the BMLDs ranged from 7.3 to 11.5 dB. The TMR for 50% correct Mandarin tone recognitions differed significantly (p<0.001) between the S0N0 and SπN0 conditions, at -13.4 and -18.0 dB, respectively, with a mean BILD of 4.6 dB. The study showed that the thresholds of Mandarin tone detection and recognition in the presence of speech-spectrum noise are improved when phase inversion is applied to the target speech. The average BILDs of Mandarin tones are smaller than the average BMLDs of Mandarin tones.
双耳听觉涉及利用与到达两只耳朵的信号差异相关的信息，并且它可以使在嘈杂环境中检测和识别信号变得更容易。在实验室研究中，这种双耳听觉现象被量化为双耳掩蔽级差（BMLD）。普通话是最常用的语言之一，但基于普通话声调的BMLD或双耳强度差（BILD）尚无公开数值。因此，本研究调查了普通话声调的BMLD和BILD。普通话声调检测的BMLD是根据在语音频谱噪声存在的情况下，浊音元音/i/（即/i1/、/i2/、/i3/和/i4/）和/u/（即/u1/、/u2/、/u3/和/u4/）的四个声调在双耳同相（S0N0）和双耳反相（SπN0）呈现时的检测阈值差异来测量的。语音频谱噪声中普通话声调识别的BILD被确定为在S0N0和SπN0条件下50%正确声调识别所需的目标与掩蔽比（TMR）的差异。/i/和/u/的四个声调的检测阈值在S0N0和SπN0条件之间有显著差异（p<0.001）。普通话声调的平均检测阈值在SπN0条件下均低于S0N0条件，BMLD范围为7.3至11.5 dB。50%正确普通话声调识别的TMR在S0N0和SπN0条件之间有显著差异（p<0.001），分别为-13.4 dB和-18.0 dB，平均BILD为4.6 dB。研究表明，当对目标语音应用相位反转时，在语音频谱噪声存在的情况下，普通话声调检测和识别的阈值会得到改善。普通话声调的平均BILD小于普通话声调的平均BMLD。
