In natural continuous speech, speakers articulate at both Inspiratory Reserve Volume (IRV) and Expiratory Reserve Volume (ERV), commonly referred to as high and low lung volumes respectively. This paper aims to investigate whether glottal parameters of a speaker portray the variations in their corresponding lung volumes during vowel articulation. It also seeks to ascertain the extent to which each parameter conveys this information. To capture these parameters, we employ two Glottal Closure Instant (GCI) detection algorithms: Speech Event Detection using the Residual Excitation And a Mean-based Signal (SEDREAMS) and Single Frequency Filtering (SFF). These observations are made while participants are engaged in continuous reading of a phonetically balanced paragraph over a span of 2–3 min. Our research delves into examining how lung volume influences glottal parameters among 94 speakers as they articulate vowels with the first formant exceeding 700 Hz and the second formant below 2500 Hz. Among a vector of 24 glottal parameters, three consistently show higher values during vowel articulation at high lung volume. Leveraging these three parameters, we can accurately differentiate between high and low lung volume in vowel articulation among 77% of the speakers.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

The Effect of Lung Volume on Glottal Parameters: An Empirical Study

  • Gauri Deshpande,
  • Björn W. Schuller

摘要

In natural continuous speech, speakers articulate at both Inspiratory Reserve Volume (IRV) and Expiratory Reserve Volume (ERV), commonly referred to as high and low lung volumes respectively. This paper aims to investigate whether glottal parameters of a speaker portray the variations in their corresponding lung volumes during vowel articulation. It also seeks to ascertain the extent to which each parameter conveys this information. To capture these parameters, we employ two Glottal Closure Instant (GCI) detection algorithms: Speech Event Detection using the Residual Excitation And a Mean-based Signal (SEDREAMS) and Single Frequency Filtering (SFF). These observations are made while participants are engaged in continuous reading of a phonetically balanced paragraph over a span of 2–3 min. Our research delves into examining how lung volume influences glottal parameters among 94 speakers as they articulate vowels with the first formant exceeding 700 Hz and the second formant below 2500 Hz. Among a vector of 24 glottal parameters, three consistently show higher values during vowel articulation at high lung volume. Leveraging these three parameters, we can accurately differentiate between high and low lung volume in vowel articulation among 77% of the speakers.