An Interspeech 2026 data challenge on detecting voice disorders from over 6,000 hours of neck-surface vibrations recorded in daily life
The challenge ended on February 20, 2026!
The NeckVibe Challenge at Interspeech 2026 conference invites researchers to submit papers on developing machine-learning models to detect two common voice-disorder subtypes: phonotraumatic (PVH) and nonphonotraumatic (NPVH) vocal hyperfunction. Participants will, for the first time, access a large-scale ambulatory dataset of over 6,000 hours of neck-surface vibration recordings collected using a neck-attached accelerometer (ACC) connected to a smartphone-based monitoring system during daily life. By transforming neck-surface vibrations into voice biomarkers, the challenge brings clinical insight into daily life, enabling remote, continuous monitoring and personalized, accessible voice care.
Key characteristics of the challenge:
Uses neck-surface vibration data only (accelerometer sensor is robust to noise and does not capture intelligible speech, which preserves privacy)
Provides over 6,000 hours of real-world voice monitoring
Includes patients and matched controls, with two tasks: PVH vs non-PVH and NPVH vs non-NPVH
Encourages clinically interpretable models with strong generalization
Acknowledgments: The NeckVibe Challenge is hosted by the Vocal Hyperfunction Clinical Research Center (VHCRC), which is supported by funding from the National Institutes of Health National Institute on Deafness and Other Communication Disorders. Website content is solely the responsibility of the investigators and does not necessarily represent the official views of the National Institutes of Health.
First-ever dataset with over 6,000 hours of neck-surface vibration data collected in daily life.
Train models on two of the most common voice disorder subtypes (PVH & NPVH) using clinically validated data.
Shape next-generation vocal health, turning real-world data into tools for early detection and individualized care.
The challenge ended on February 20, 2026
Explore the pages below to learn more about the challenge
