Identification of vocal tract state before and after swallowing using acoustic features

Identification of vocal tract state before and after swallowing using acoustic features
复制标题

利用声学特征识别吞咽前后的声道状态

DOI:
10.1109/gcce56475.2022.10014393
复制
发表时间:
2023
期刊:
Proc. 2022 IEEE 11th Global Conference on Consumer Electronics
影响因子:
--
通讯作者:
Shingo Kuroiwa
Shingo Kuroiwa
中科院分区:
--
文献类型:
--
作者:
Aoi Sugita;Masafumi Nishida;Masafumi Nishimura;Yasuo Horiuchi;Shingo Kuroiwa

文献摘要

参考文献

相似文献

吞咽后残留在咽部的食物是误吸的原因之一。为了检测咽部的残留食物,我们研究了一种非侵入性的方法,该方法仅从声音来估计包括咽部在内的声道的状况。在这项研究中,我们检查了四种声道状态的识别:吞咽前,干吞咽(吞咽唾液)后,吞咽水后,吞咽酸奶后。为了识别四种声道状态,我们采用了对数功率谱包络和支持向量机(SVM)。除了通常的四类识别对应的四个声道状态,两种方法被认为是提高对扬声器之间的变化的鲁棒性:1)的方法,同时确定声道状态和扬声器(4类扬声器的数量)和2)扬声器归一化的方法。实验结果表明,在说话人封闭条件下,不进行说话人归一化的4状态16说话人识别方法具有最高的识别率。另一方面,在说话人开放的条件下,准确率仍然很低,虽然归一化提高了识别率。
Residual food in the pharynx after swallowing is one of the causes of aspiration. To detect residual food in the pharynx, we investigate a noninvasive method that estimates the condition of the vocal tract, which includes the pharynx, only from voice. In this study, we examined the identification of four vocal tract states: before swallowing, after dry swallowing (swallowing saliva), after water swallowing, and after drinkable yogurt swallowing. To identify the four vocal tract states, we employed a log-power spectral envelope and a support vector machine (SVM). In addition to the usual four-class identification corresponding to the four vocal tract states, two methods were considered to improve robustness against variation among speakers: 1) a method that simultaneously determines the vocal tract state and the speaker (4-classnumber of speakers) and 2) a method of speaker normalization. Experimental results indicated that the proposed method (4 states16 speakers) without speaker normalization had the highest identification rate under the speaker-closed condition. On the other hand, under the speaker-open condition, the accuracy was still low, although normalization improved the identification rate.
DOI: 10.1016/j.anl.2017.10.007
发表时间: 2018-08-01
期刊: AURIS NASUS LARYNX
影响因子: 1.7
作者:
Kang, Young Ae;Kim, Jaeock;Koo, Bon Seok
通讯作者: Koo, Bon Seok
通过呼吸和吞咽声音分析进行无声误吸检测
DOI: 10.1109/embc.2012.6346496
发表时间: 2012
期刊: 2012 Annual International Conference of the IEEE Engineering in Medicine and Biology Society
影响因子: --
作者:
Samaneh Sarraf Shirazi;Z. Moussavi
通讯作者: Z. Moussavi
吞咽前、吞咽中和吞咽后误吸的修订分类及其可靠性
DOI: 10.11336/jjcrs.7.51
发表时间: 2016
期刊: Japanese Journal of Comprehensive Rehabilitation Science
影响因子: --
作者:
Takashi Tanaka;H. Kagaya;Yasunori Ozeki;W. Fujii;Takatoshi Iida;S. Shibata;Saki Tomita;Yuriko Ishiguro;K. Ota;E. Saitoh
通讯作者: E. Saitoh