Toward ambulatory monitoring of vocal behavior at the physiological level using deep ensembles and Bayesian neural networks

2025/11/01 by Zhaoyan Zhang · 1 voice
Computer Science · Medicine · Psychology · #Speech Recognition and Synthesis #Stuttering Research and Treatment #Voice and Speech Disorders

paper · pdf · doi:10.1121/10.0039842

openalex publication_date 2025/11/01 · openalex created_date 2025/11/11 · openalex updated_date 2026/08/01

Abstract

Currently, diagnosis of voice disorders is often made when patients visit the clinic, by which time speakers already experience vocal difficulties. The goal of this study was to develop a voice inversion system that predicts how speakers modulate vocal physiology from the produced voice, toward early detection of unhealthy vocal behavior. Two neural networks, a Bayesian neural network and a deep ensemble of neural networks, were developed that predict changes in vocal physiological parameters and their confidence intervals. Comparison to human data showed that both networks were able to predict meaningful differences in vocal behavior across subjects, demonstrating their potential toward ambulatory monitoring of vocal behavior at the physiological level.

Citations

• Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles

Cited by

• Gradient boundaries through confidence intervals for forced alignment estimates using model ensembles

Discussions

• JASA-EL EDITOR'S PICK Toward ambulatory monitoring of vocal behavior at the physiological level using deep ensembles and Bayesian neural networks https://doi.org/10.1121/10.0039842 [bsky, 1 points, 0 comments]

Related

• Emotion Recognition of the Singing Voice: Toward a Real-Time Analysis Tool for Singers
• Uncovering Voice Misuse Using Symbolic Mismatch
• Voice Pathology Detection Using Phonation
• Deep Neural Network for Automatic Assessment of Dysphonia
• Detecting Vocal Fatigue with Neural Embeddings