Automatic speech and singing classification in ambulatory recordings for normal and disordered voices

• Main Authors: Capobianco, S. (Author), Hillman, R.E (Author), Marks, K.L (Author), Mehta, D.D (Author), Ortiz, A.J (Author), Toles, L.E (Author), Van Stan, J.H (Author)
• Format: Article
• Language: English
• Published: Acoustical Society of America 2019
• Subjects: adult; Adult; ambulatory monitoring; article; Automatic speech; Classification (of information); Classification accuracy; Disordered voice; female; Female; Fundamental frequencies; Ground truth data; Healthy controls; human; human experiment; Humans; Logistic regressions; Monitoring, Ambulatory; pathophysiology; physiology; procedures; singing; Singing; speech; Speech; vocal cord; Vocal Cords; Vocal hyperfunction; voice; Voice; voice disorder; Voice Disorders
• Online Access: View Fulltext in Publisher
• ISBN: 00014966 (ISSN)
• DOI: 10.1121/1.5115804

Ambulatory voice monitoring is a promising tool for investigating phonotraumatic vocal hyperfunction (PVH), associated with the development of vocal fold lesions. Since many patients with PVH are professional vocalists, a classifier was developed to better understand phonatory mechanisms during speech and singing. Twenty singers with PVH and 20 matched healthy controls were monitored with a neck-surface accelerometer-based ambulatory voice monitor. An expert-labeled ground truth data set was used to train a logistic regression on 15 subject-pairs with fundamental frequency and autocorrelation peak amplitude as input features. Overall classification accuracy of 94.2% was achieved on the held-out test set. © 2019 Acoustical Society of America.