Suppr超能文献

随着通道数量增加,普通话词汇声调感知能力持续改善:一项人工耳蜗模拟研究

Continuous improvement in Mandarin lexical tone perception as the number of channels increased: a simulation study of cochlear implant.

作者信息

Lin Yung-Song, Lee Fei-Peng, Huang I-Shun, Peng Shu-Chen

机构信息

Department of Otolaryngology, Taipei Medical University, Chi Mei Medical Center, Tainan city, Taiwan, ROC.

出版信息

Acta Otolaryngol. 2007 May;127(5):505-14. doi: 10.1080/00016480600951434.

Abstract

CONCLUSION

With reference to English phoneme recognition, where performance usually does not improve after six or eight channels in cochlear implants (CIs), increasing total channel numbers continuously improved perception of Mandarin tones.

OBJECTIVE

To test our hypothesis that current CI strategies might be modified to improve Mandarin lexical tonal perception.

MATERIALS AND METHODS

Lexical tonal perception tests using 48 monosyllables in Mandarin Chinese were conducted in 32 native Mandarin speakers with normal hearing. The performance of tonal perception was compared among the controlled factors, which were total channel number, number of channels allocated to the F0 spectrum, and whether there were spectral shifts in the electrode configuration. The experimental condition that preserves fine structure was used as a comparison.

RESULTS

The signal processing strategy using 16 channels--which is technically possible with current CI devices--produced better tonal perception than those using 12 or 8 channels. Increasing the number of fundamental channels did not improve tonal perception, and spectral shifts did not change tonal perception. An experimental condition (FiC12) that preserves the fine structure produced significantly better overall scores for tone perception than other experimental conditions with envelope strategies.

摘要

结论

关于英文音素识别,人工耳蜗(CI)通常在使用6个或8个通道后性能不再提升,但增加通道总数能持续改善对汉语声调的感知。

目的

验证我们的假设,即当前的CI策略或许可进行调整以改善汉语词汇声调感知。

材料与方法

对32名母语为汉语且听力正常的受试者进行了汉语词汇声调感知测试,使用48个汉语单音节。在总通道数、分配给基频(F0)频谱的通道数以及电极配置中是否存在频谱偏移这些控制因素之间比较声调感知性能。将保留精细结构的实验条件用作对照。

结果

使用16个通道的信号处理策略(这在当前的CI设备上技术上是可行的)比使用12个或8个通道的策略产生了更好的声调感知。增加基频通道数并未改善声调感知,频谱偏移也未改变声调感知。与采用包络策略的其他实验条件相比,保留精细结构的实验条件(FiC12)在声调感知方面产生了显著更高的总体分数。

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验