基于气息音的嗓音分类研究

刘阳; 李进让

您当前的位置：

首页 >

文章列表页 >

基于气息音的嗓音分类研究

嗓音康复治疗专题 | 更新时间：2026-10-28

- 基于气息音的嗓音分类研究
- Voice Classification Based on Breathiness Features
- 中国听力语言康复科学杂志 2022年20卷第1期页码：8-11
- 作者机构：
  
  解放军总医院第六医学中心耳鼻咽喉头颈外科医学部咽喉嗓音外科
- 作者简介：
- 基金信息：
- DOI：
  中图分类号： R767.92
- 纸质出版：2022
- 稿件说明：
移动端阅览
刘阳, 李进让. 基于气息音的嗓音分类研究[J]. 中国听力语言康复科学杂志, 2022,20(1):8-11.

刘阳, 李进让. Voice Classification Based on Breathiness Features[J]. 2022, 20(1): 8-11.
刘阳, 李进让. 基于气息音的嗓音分类研究[J]. 中国听力语言康复科学杂志, 2022,20(1):8-11. DOI：

刘阳, 李进让. Voice Classification Based on Breathiness Features[J]. 2022, 20(1): 8-11. DOI：

摘要

目的以量化的气息音为特征，对嗓音疾病进行分类预测。方法对声带麻痹、器质性嗓音疾病、功能性嗓音疾病患者及正常受试者各30例的嗓音进行音频片段抽取，去除音频头尾各0.25秒，以100毫秒为1帧，每例抽取20帧，将每帧保存为独立语音片段。使用voicesauce对语音片段进行分析，提取气息音特征参数（H1-H2

H1-A1

H1-A3

CPP）数值，清洗后建立数据集。使用随机森林方法分别在不同决策树规模和不同特征数量下建立分类器模型，并检验分类正确率。结果不同决策树规模下（100棵和200棵），单一特征和组合特征的准确率无显著区别。相同决策树规模下，单一特征的分类准确率分别为H1-H2:27.56%

H1-A1:30.16%

H1-A3:29.58%

CPP:38.24%；组合特征（H1-H2&H1-A1&H1-A3&CPP）的分类准确率为71.14%。在组合特征中，最重要的特征是CPP，约占28.06%。结论通过客观声学测量获得气息音组合特征可作为区分功能性嗓音疾病、器质性嗓音疾病、声带麻痹和正常嗓音的分类特征，可以获得较好的分类效果。

Abstract

Objective The purpose of this study is to classify 4 different voice types with quantified breathiness features.Methods 30 voice recordings were selected in each of the four groups（vocal cord paralysis patients

organic voice disease patients

functional voice disease patients

and normal subjects）. After removing 0.25 seconds for each end of the audios

100 milliseconds was randomly taken as one frame

20 frames were extracted from every case

and each frame was saved as an independent wav file.Voice sauce was used to analyze each wave file

extract the values of breathiness characteristic parameters（H1-H2

H1-A1

H1-A3

CPP）

and set up a new dataset after cleaning. The Random Forest was used to build classifier models with different decision tree sizes and different feature numbers

the accuracy and feature importance were calculated. Results Under different decision tree sizes（100 and 200 trees）

there was no significant difference in the accuracy of single feature and combination feature. Under the same decision tree scale

the accuracy of single feature was 27.56% for H1-H2

30.16% for H1-A1

29.58% for H1-A3 and 38.24% for CPP. The accuracy of combination features（H1-H2-H1-A1-H1-A3 ＆ CPP） was 71.14%. Among the combination characteristics

CPP was the most important

accounting for 28.06%. Conclusion The combination feature of breathiness can be used to distinguish functional voice diseases

organic voice diseases

vocal cord paralysis and normal voice

and can achieve a good classification accuracy.

关键词

Keywords

references

周林灿.病理嗓音的声学分析和合成[D].华东师范大学,2011(09).

薛隆基.病理嗓音的递归量化分析及分类研究[D].苏州大学,2020(02).

谢思敏.基于深度神经网络的病理嗓音分类研究[D].武汉理工大学,2017(02).

Yolanda D. Heman-Ackah,Robert T. Sataloff,Griet Laureyns,Deborah Lurie,Deirdre D. Michael,Reinhardt Heuer,Adam Rubin,Robert Eller,Swapna Chandran,Mona Abaza,Karen Lyons,Venu Divi,Joanna Lott,Jennifer Johnson,James Hillenbrand.Quantifying the Cepstral Peak Prominence, a Measure of Dysphonia[J].Journal of Voice,2014.

Katia Nemr,Marcia Simões-Zenari,Gislaine Ferro Cordeiro,Domingos Tsuji,Allex Itar Ogawa,Maysa Tibério Ubrig,Márcia Helena Moreira Menezes.GRBAS and Cape-V Scales: High Reliability and Consensus When Applied at Different Times[J].Journal of Voice,2012.

Kreiman Jody,Shue Yen-Liang,Chen Gang,Iseli Markus,Gerratt Bruce R,Neubauer Juergen,Alwan Abeer.Variability in the relationships among voice quality, harmonic amplitudes, open quotient, and glottal area waveform shape in sustained phonation.[J].The Journal of the Acoustical Society of America,2012.

Samlan Robin A,Story Brad H.Relation of structural and vibratory kinematics of the vocal folds to two acoustic measures of breathy voice based on computational modeling.[J].Journal of speech, language, and hearing research : JSLHR,2011.

Jody Kreiman,Bruce R. Gerratt,Sameer ud Dowla Khan.Effects of native language on perception of voice quality[J].Journal of Phonetics,2010.

Christian T. DiCanio.The phonetics of register in Takhian Thong Chong[J].Journal of the International Phonetic Association,2009.

Helen M. Hanson,Kenneth N. Stevens,Hong-Kwang Jeff Kuo,Marilyn Y. Chen,Janet Slifka.Towards models of phonation[J].Journal of Phonetics,2001.

Melissa A. Epstein.A comparison of linguistic and pathological breathiness using the LF model[J].The Journal of the Acoustical Society of America,2000.

Holmberg E B,Hillman R E,Perkell J S,Guiod P C,Goldman S L.Comparisons among aerodynamic, electroglottographic, and acoustic spectral measures of female voice.[J].Journal of speech and hearing research,1995(6).

Smith M E,Berke G S,Gerratt B R,Kreiman J.Laryngeal paralyses: theoretical considerations and effects on laryngeal vibration.[J].Journal of speech and hearing research,1992(3).

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

内耳畸形的分类和干预

基于社会故事法改善孤独症谱系障碍儿童同伴交往主动性个案研究

人工耳蜗蓝牙直连与传统声场下言语测听效果对比

孤独症谱系障碍学生社交能力发展支持体系建构个案研究——基于ICF评估系统的融合教育实践

孤独症谱系障碍儿童融合教育实施困境与应对策略研究