利用机器学习从语音声学估计声带生理机能。
Estimation of vocal fold physiology from voice acoustics using machine learning.
作者信息
Zhang Zhaoyan
机构信息
Department of Head and Neck Surgery, University of California, Los Angeles, 31-24 Rehab Center, 1000 Veteran Avenue, Los Angeles, California 90095-1794,
出版信息
J Acoust Soc Am. 2020 Mar;147(3):EL264. doi: 10.1121/10.0000927.
Abstract
The goal of this study is to estimate vocal fold geometry, stiffness, position, and subglottal pressure from voice acoustics, toward clinical and other voice technology applications. Unlike previous voice inversion research that often uses lumped-element models of phonation, this study explores the feasibility of voice inversion using data generated from a three-dimensional voice production model. Neural networks are trained to estimate vocal fold properties and subglottal pressure from voice features extracted from the simulation data. Results show reasonably good estimation accuracy, particularly for vocal fold properties with a consistent global effect on voice production, and reasonable agreement with excised human larynx experiment.
摘要
本研究的目标是从语音声学估计声带的几何形状、硬度、位置和声门下压力，以用于临床和其他语音技术应用。与以往常使用发声集总元件模型的语音反演研究不同，本研究探索了使用三维语音产生模型生成的数据进行语音反演的可行性。训练神经网络从模拟数据中提取的语音特征来估计声带特性和声门下压力。结果显示出相当不错的估计精度，特别是对于对语音产生具有一致全局影响的声带特性，并且与切除的人体喉部实验结果具有合理的一致性。
相似文献
1
Estimation of vocal fold physiology from voice acoustics using machine learning.利用机器学习从语音声学估计声带生理机能。
J Acoust Soc Am. 2020 Mar;147(3):EL264. doi: 10.1121/10.0000927.
2
Voice Feature Selection to Improve Performance of Machine Learning Models for Voice Production Inversion.语音特征选择以提高语音产生反转机器学习模型的性能。
J Voice. 2023 Jul;37(4):479-485. doi: 10.1016/j.jvoice.2021.03.004. Epub 2021 Apr 11.
3
Cause-effect relationship between vocal fold physiology and voice production in a three-dimensional phonation model.三维发声模型中声带生理学与发声之间的因果关系。
J Acoust Soc Am. 2016 Apr;139(4):1493. doi: 10.1121/1.4944754.
4
Effect of vocal fold stiffness on voice production in a three-dimensional body-cover phonation model.声带僵硬对三维体罩发声模型中发声的影响。
J Acoust Soc Am. 2017 Oct;142(4):2311. doi: 10.1121/1.5008497.
5
Synthetic, multi-layer, self-oscillating vocal fold model fabrication.合成多层自振荡声带模型的制作。
J Vis Exp. 2011 Dec 2(58):3498. doi: 10.3791/3498.
6
The Effect of Vocal Fold Inferior Surface Hypertrophy on Voice Function in Excised Canine Larynges.声带下表面肥厚对犬离体喉嗓音功能的影响。
J Voice. 2018 Jul;32(4):396-402. doi: 10.1016/j.jvoice.2017.06.013. Epub 2017 Aug 18.
7
Experimental study of vocal-ventricular fold oscillations in voice production.发声时声襞-室带的运动实验研究。
J Acoust Soc Am. 2021 Jan;149(1):271. doi: 10.1121/10.0003211.
8
Estimation of Subglottal Pressure, Vocal Fold Collision Pressure, and Intrinsic Laryngeal Muscle Activation From Neck-Surface Vibration Using a Neural Network Framework and a Voice Production Model.使用神经网络框架和语音产生模型从颈部表面振动估计声门下压力、声带碰撞压力和喉内肌激活。
Front Physiol. 2021 Sep 1;12:732244. doi: 10.3389/fphys.2021.732244. eCollection 2021.
9
Subglottal pressure oscillations in anechoic and resonant conditions and their influence on excised larynx phonations.无声和共鸣条件下的声门下压力振荡及其对离体喉发声的影响。
Sci Rep. 2021 Jan 8;11(1):28. doi: 10.1038/s41598-020-79265-3.
10
Indirect assessment of the contribution of subglottal air pressure and vocal-fold tension to changes of fundamental frequency in English.英语中声门下气压和声带张力对基频变化贡献的间接评估。
J Acoust Soc Am. 1978 Jul;64(1):65-80. doi: 10.1121/1.381957.
引用本文的文献
1
Estimation of Physiological Vocal Features from Neck Surface Acceleration Signals Using Probabilistic Bayesian Neural Networks.使用概率贝叶斯神经网络从颈部表面加速度信号估计生理发声特征。
IEEE Trans Audio Speech Lang Process (2025). 2025;33:1576-1589. doi: 10.1109/taslpro.2025.3552938. Epub 2025 Apr 18.
2
Early fiber development in human vocal folds: An in vitro pilot study.人类声带早期纤维发育：一项体外初步研究。
JASA Express Lett. 2025 Jul 1;5(7). doi: 10.1121/10.0037106.
3
Subject-Specific Modeling by Domain Adaptation for the Estimation of Subglottal Pressure from Neck-Surface Acceleration Signals.基于域适应的特定对象建模，用于从颈部表面加速度信号估计声门下压力
Biomed Signal Process Control. 2025 Aug;106. doi: 10.1016/j.bspc.2025.107681. Epub 2025 Feb 26.
4
Characterization of upper esophageal sphincter pressures relative to vocal acoustics.相对于嗓音声学的食管上括约肌压力特征
J Appl Physiol (1985). 2025 Jan 1;138(1):203-212. doi: 10.1152/japplphysiol.00385.2024. Epub 2024 Dec 6.
5
Deep Learning for Neuromuscular Control of Vocal Source for Voice Production.用于语音产生中声源神经肌肉控制的深度学习
Appl Sci (Basel). 2024 Jan;14(2). doi: 10.3390/app14020769. Epub 2024 Jan 16.
6
Neural network-based estimation of biomechanical vocal fold parameters.基于神经网络的生物力学声带参数估计
Front Physiol. 2024 Feb 21;15:1282574. doi: 10.3389/fphys.2024.1282574. eCollection 2024.
7
An introduction to machine learning and generative artificial intelligence for otolaryngologists-head and neck surgeons: a narrative review.耳鼻喉科-头颈外科医师的机器学习和生成式人工智能入门：叙述性综述。
Eur Arch Otorhinolaryngol. 2024 May;281(5):2723-2731. doi: 10.1007/s00405-024-08512-4. Epub 2024 Feb 23.
8
Toward Generalizable Machine Learning Models in Speech, Language, and Hearing Sciences: Estimating Sample Size and Reducing Overfitting.迈向语音、语言和听力科学中的通用机器学习模型：估计样本量并减少过拟合。
J Speech Lang Hear Res. 2024 Mar 11;67(3):753-781. doi: 10.1044/2023_JSLHR-23-00273. Epub 2024 Feb 22.
9
Deep learning in voice analysis for diagnosing vocal cord pathologies: a systematic review.深度学习在声纹分析诊断声带病变中的应用：系统综述。
Eur Arch Otorhinolaryngol. 2024 Feb;281(2):863-871. doi: 10.1007/s00405-023-08362-6. Epub 2023 Dec 13.
10
3D VOSNet: Segmentation of endoscopic images of the larynx with subsequent generation of indicators.3D VOSNet：用于分割喉部内窥镜图像并随后生成指标
Heliyon. 2023 Mar 3;9(3):e14242. doi: 10.1016/j.heliyon.2023.e14242. eCollection 2023 Mar.
本文引用的文献
1
Bayesian Inference of Vocal Fold Material Properties from Glottal Area Waveforms Using a 2D Finite Element Model.使用二维有限元模型从声门面积波形进行声带材料特性的贝叶斯推断
Appl Sci (Basel). 2019 Jul 1;9(13). doi: 10.3390/app9132735. Epub 2019 Jul 6.
2
Voice production in a MRI-based subject-specific vocal fold model with parametrically controlled medial surface shape.基于 MRI 的特定个体声带模型中的语音产生，具有参数控制的中表面形状。
J Acoust Soc Am. 2019 Dec;146(6):4190. doi: 10.1121/1.5134784.
3
Machine learning in acoustics: Theory and applications.机器学习在声学中的理论与应用。
J Acoust Soc Am. 2019 Nov;146(5):3590. doi: 10.1121/1.5133944.
4
Laryngeal Pressure Estimation With a Recurrent Neural Network.基于循环神经网络的喉压估计
IEEE J Transl Eng Health Med. 2018 Dec 27;7:2000111. doi: 10.1109/JTEHM.2018.2886021. eCollection 2019.
5
Vocal instabilities in a three-dimensional body-cover phonation model.三维体罩发声模型中的声不稳定现象。
J Acoust Soc Am. 2018 Sep;144(3):1216. doi: 10.1121/1.5053116.
6
Physical parameter estimation from porcine ex vivo vocal fold dynamics in an inverse problem framework.基于反问题框架的猪离体声带动力学的物理参数估计。
Biomech Model Mechanobiol. 2018 Jun;17(3):777-792. doi: 10.1007/s10237-017-0992-5. Epub 2017 Dec 11.
7
Effect of vocal fold stiffness on voice production in a three-dimensional body-cover phonation model.声带僵硬对三维体罩发声模型中发声的影响。
J Acoust Soc Am. 2017 Oct;142(4):2311. doi: 10.1121/1.5008497.
8
Mechanics of human voice production and control.人类发声与控制的机制。
J Acoust Soc Am. 2016 Oct;140(4):2614. doi: 10.1121/1.4964509.
9
Non-stationary Bayesian estimation of parameters from a body cover model of the vocal folds.基于声带身体覆盖模型的参数非平稳贝叶斯估计。
J Acoust Soc Am. 2016 May;139(5):2683. doi: 10.1121/1.4948755.
10
Toward a unified theory of voice production and perception.迈向语音产生与感知的统一理论。
Loquens. 2014 Jan;1(1). doi: 10.3989/loquens.2014.009.
Zotero 插件