上海交通大学学报(医学版) ›› 2017, Vol. 37 ›› Issue (11): 1565-.doi: 10.3969/j.issn.1674-8115.2017.11.020

• 综述 • 上一篇    下一篇

不同频段时域包络信息在言语识别中相对重要性的研究进展

郭洋,孙渊渊,夏俍,冯艳梅   

  1. 上海交通大学附属第六人民医院耳鼻咽喉头颈外科,上海交通大学医学院耳鼻咽喉科研究所,上海 200233
  • 出版日期:2017-11-28 发布日期:2018-01-10
  • 通讯作者: 冯艳梅,电子信箱:feng.yanmei@126.com
  • 作者简介:郭洋(1994—),男,硕士生;电子信箱:guoyangls@126.com
  • 基金资助:
    上海市教育委员会高峰高原学科建设计划(20152526);国家自然科学基金面上项目(81771015);促进市级医院临床技能与临床创新三年行动计 划(16CR4027A)

Research progresses of relative importance of temporal envelope cues in different frequency regions#br#

GUO Yang, SUN Yuan-yuan, XIA Liang, FENG Yan-mei   

  1. Department of Otolaryngology, Shanghai Sixth People’s Hospital, Institute of Otolaryngology, Shanghai Jiao Tong University, Shanghai 200233, China
  • Online:2017-11-28 Published:2018-01-10
  • Supported by:
    Shanghai Municipal Education Commission—Gaofeng Clinical Medicine Grant Support,20152526;National Natural Science Foundation of China, 81771015;Three-year Action Program on Promotion of Clinical Skills and Clinical Innovation for Municipal Hospitals,16CR4027A

摘要: 言语信号中的时域信息根据其随时间变化的速率可以分为时域包络信息、周期性波动信息和时域精细结构信息。时域包络信 息在言语识别中占有重要地位,是如今大部分人工耳蜗可以传递给其使用者的言语信息。研究表明不同频段的时域包络信息在言语识 别中的作用并不相同。受测试材料、研究方法、聆听环境、提取时域包络信息的参数的影响,不同频段时域包络信息的相对重要性也 会随之改变。文章主要就不同频段时域包络信息在言语识别中相对重要性的研究方法以及各种方法的优缺点和研究结果进行综述,并 初步探讨了非声调语言与声调语言不同频段时域包络信息在言语识别中相对重要性差异的原因。

关键词: 时域包络, 频段, 相对重要性, 声调语言

Abstract:

Based on the dominant fluctuation rates, the speech information in temporal domain could be divided into temporal envelope, periodic fluctuation information and temporal fine structure. Temporal envelope cues are essential for speech recognition, which could be transmitted to cochlear implanters by cochlear imlpants. The roles of temporal envelope cues from various frequency regions in speech recognition are diverse. Influenced by the testing materials, research methods, listening backgrounds and the parameters used to extract temporal envelope, the relative weights of temporal envelope across frequency regions would change accordingly. The research methods as well as their advantages or disadvantages and research results of relative weights of temporal envelope cues in different frequency regions are reviewed, and the possible reasons why the relative weights of temporal envelope cues in different frequency regions for non-tonal language and tonal language were different were discussed simply.

Key words: temporal envelope, frequency region, relative importance, tonal language

中图分类号: