病态嗓音的定量分析及人工神经网络识别

发布时间：2018-07-03 02:01

本文选题：嗓音声学分析 + 病态嗓音　；参考：《临床耳鼻咽喉头颈外科杂志》2017年02期

【摘要】：目的:探讨临床病态嗓音的特征及计算机自动识别病态嗓音的可行性。方法:选择129例声带息肉患者为病态嗓音组,同期选取125例社区正常嗓音人群为对照组。应用Praat软件采集分析2组病例获得相关声学参数值,包括基频微扰、振幅微扰、谐噪比、信噪比、声门噪声。采用该病态嗓音组与对照组病例作为神经网络检测的训练集和测试集。同样方法另外收集140例病态嗓音及正常嗓音数据作为验证集。应用SPSS Modeler软件进行人工神经网络建模,计算模型对病态嗓音的识别率。结果:本研究根据不同性别分组计算,病态嗓音组在基频微扰、振幅微扰、声门噪声方面数值比对照组增大(P0.05),谐噪比、信噪比方面数值比对照组减少(P0.05)。人工神经网络模型对病态嗓音的识别率为75.7%。结论:客观嗓音分析有助于病态嗓音的鉴别,人工神经网络在病态嗓音的识别上准确率较高,有很好的临床应用价值。
[Abstract]:Objective: to explore the features of clinical pathological voice and the feasibility of computer automatic recognition of pathological voice. Methods: 129 patients with vocal cord polyps were selected as pathological voice group and 125 normal voice patients as control group. Two groups of cases were collected and analyzed by Praat software to obtain relevant acoustic parameters, including fundamental frequency perturbation, amplitude perturbation, harmonic noise ratio, signal-to-noise ratio and glottic noise. The pathological voice group and the control group were used as the training set and test set for neural network detection. In the same method, 140 samples of abnormal voice and normal voice were collected as validation set. SPSS Modeler software was used to model artificial neural network, and the recognition rate of pathological voice was calculated. Results: according to different sex groups, the fundamental frequency perturbation, amplitude perturbation and glottic noise were higher in the pathological voice group than in the control group (P0.05), the harmonic noise ratio and the signal-to-noise ratio were lower than those in the control group (P0.05). The recognition rate of artificial neural network model for pathological voice is 75.775%. Conclusion: objective voice analysis is helpful to the identification of pathological voice. The accuracy of artificial neural network in the recognition of pathological voice is high and has good clinical application value.
【作者单位】：复旦大学附属上海市第五人民医院耳鼻咽喉科;上海市虹口区凉城新村街道社区卫生服务中心;
【基金】：复旦大学附属上海市第五人民医院课题(No:2010WYQJ02)
【分类号】：R767.92

【相似文献】