Zheng Xiao L, Fan Xing Y, Hou Wen S
Annu Int Conf IEEE Eng Med Biol Soc. 2016 Aug;2016:1608-1611. doi: 10.1109/EMBC.2016.7591020.
Mandarin speech has four different tones and the coding mechanism underlying tone identification still remain unclear. Here in the inferior colliculus (IC) of anesthetized guinea pigs, we recorded single neuron activities to one word with four tones using tungsten electrode. Peri-stimulus time histograms (PSTHs) and inter-spike-interval (ISI) were used to evaluate the neural response. The results showed that PSTHs grouped into frequency band reflected the spectrotemporal patterns of different tones; average population PSTHs matched envelops of different tones; and the peaks of histogram of ISIs in three time segments exhibited a displacement which reflected the profile of fundamental frequency (F0). These preliminary results suggested IC neurons could encode the spectrotemporal acoustic features of different Mandarin tones.
汉语语音有四种不同的声调，而声调识别背后的编码机制仍不清楚。在此，我们在麻醉豚鼠的下丘（IC）中，使用钨电极记录了单个神经元对一个包含四种声调的单词的活动。采用刺激周围时间直方图（PSTH）和峰峰间期（ISI）来评估神经反应。结果表明，按频段分组的PSTH反映了不同声调的频谱时间模式；平均群体PSTH与不同声调的包络相匹配；并且ISI直方图在三个时间段的峰值呈现出位移，这反映了基频（F0）的特征。这些初步结果表明，IC神经元可以编码不同汉语声调的频谱时间声学特征。
