Choosing the right rate normalization method for measurements of speech rhythm

Choosing the right rate normalization method for measurements of speech rhythm
复制标题

选择正确的速率归一化方法来测量语音节奏

DOI:
10.5167/uzh-45236
复制
发表时间:
2009
期刊:
Canadian journal of psychiatry. Revue canadienne de psychiatrie
影响因子:
--
通讯作者:
Dieter Studer
Dieter Studer
中科院分区:
--
文献类型:
--
作者:
V. Dellwo;Stephan Schmid;Michael Schwarzenbach;Dieter Studer

文献摘要

参考文献

被引文献

相似文献

与语言节奏相关的声学因素是辅音和元音的音长特征。本研究调查的影响,语音速率对这些声学相关。在实验I中,四个广泛应用的语音节奏(%V,NPVI,NPVI和rPVI)的相关性与辅音和元音间隔使用语音从五种不同的语言(捷克语,英语,法语,德语,意大利语),其特征在于高克里思的变化,在每种语言(非常慢,非常快的语音)的速率。结果发现,节奏措施的基础上辅音间隔持续时间(CXC,rPVI)与率的措施呈负相关,节奏措施的基础上声乐间隔(%V,nPVI)不受率的影响。在实验II中的有效性率归一化程序的速率相关的措施,RPVI,通过相关这些措施与语音速率之前和之后的归一化使用相同的语音数据在实验1中进行了测试。通过对辅音间隔持续时间进行数学变换来对PVC进行归一化,并且通过先前提出的用于对nPVI进行归一化的方法来对rPVI进行归一化。结果发现,可以使用建议的心率归一化程序有效地归一化心率对RPVI和RPVC的影响。在实验III中,它被测试是否率归一化措施的语音节奏支持的印象,一些语言可以根据其听觉节奏特征(如重音和音节定时)进行分类。仅在速率标准化rPVI中发现了对此的有力支持,这是标准化CFDIC显示混合结果的原因。它的结论是,CNODC是不太适合的节奏测量,旨在分离不同的节奏类的语言。
Some acoustic correlates of language rhythm are durational characteristics of consonants and vowels. The present study investigates the influence of speech rate on these acoustic correlates. In experiment I four widely applied correlates of speech rhythm (%V, ∆C, nPVI and rPVI) were correlated with the rate of consonantal and vocalic intervals using speech from five different languages (Czech, English, French, German, Italian) that was characterized by high tempo variability within each language (very slow to very fast produced speech). It was found that rhythm measures based on consonantal interval durations (∆C, rPVI) correlate negatively with rate measures and that rhythm measures based on vocalic intervals (%V, nPVI) are not influenced by rate. In experiment II the effectiveness of rate normalization procedures on the rate dependent measures, ∆C and rPVI, was tested by correlating these measures with speech rate before and after normalization using the same speech data as in Experiment 1. ∆C was normalized by logarithmically transforming the consonantal interval durations and rPVI was normalized by previously proposed ways for the normalization of nPVI. It was found that rate effects on ∆C and rPVI could be normalized for effectively using the suggested rate normalization procedures. In Experiment III it was tested whether rate normalized measures of speech rhythm support the impression that some languages can be categorized according to their auditory rhythmic characteristics (e.g. stressand syllable-timing). Strong support for this was only found for the rate normalized rPVI why the normalized ∆C revealed mixed results. It was concluded that ∆C is less appropriate for rhythmic measurements that aim to separate languages of different rhythmic classes.
DOI: 10.1111/j.1467-7687.2005.00390.x
发表时间: 2005
影响因子: 3.7
作者:
R. Tincoff;M. Hauser;Fritz Tsao;Geertrui M. Spaepen;F. Ramus;J. Mehler
通讯作者: R. Tincoff;M. Hauser;Fritz Tsao;Geertrui M. Spaepen;F. Ramus;J. Mehler
语音节奏的低频傅立叶分析。
DOI: 10.1121/1.2947626
发表时间: 2008
期刊: The Journal of the Acoustical Society of America
影响因子: --
作者:
Tilsen,Sam;Johnson,Keith
通讯作者: Johnson,Keith
DOI: 10.1126/science.288.5464.349
发表时间: 2000-04-14
期刊: SCIENCE
影响因子: 56.9
作者:
Ramus, F;Hauser, MD;Mehler, J
通讯作者: Mehler, J