Challenge Overview

Background

Vocal hyperfunction (VH)—characterized by excessive or unbalanced laryngeal muscle activity during phonation—is a leading cause of voice disorders. It occurs in two main forms: phonotraumatic vocal hyperfunction (PVH), typically associated with tissue trauma and benign vocal fold lesions, and nonphonotraumatic vocal hyperfunction (NPVH), marked by chronic vocal fatigue and degraded voice quality without structural lesions.

Although VH is common, its etiology and persistence are still not well understood. Daily voice use is believed to play a key role, but brief clinic-based recordings capture only short snapshots of behavior and often fail to represent how people use their voices in real communicative and environmental contexts.

Why Ambulatory Neck-Surface Vibration Monitoring?

To address the limitations of clinic-based assessments, wearable systems have been developed to monitor voice use during daily life. Neck-mounted accelerometers (ACCs) capture neck-surface vibrations that reflect vocal behavior while being robust to background noise and preserving privacy compared with traditional microphones.

These systems have enabled continuous tracking of clinically interpretable voice measures in real-world settings. However, most devices can only store frame-based averages of basic measures rather than the full accelerometer signal, limiting the range of features and models that can be explored.

Aim of the NeckVibe Challenge

Our group developed a smartphone-based ambulatory voice monitor capable of recording the raw neck-surface accelerometer signal during everyday activities. This system supports continuous, real-world voice monitoring and the extraction of a wide variety of voice measures beyond simple averages.

The NeckVibe Challenge uses this technology to share a unique ambulatory dataset and invite the community to develop robust, interpretable machine learning models that:

• Detect PVH and NPVH from neck-surface vibration data

• Link real-world voice behavior to underlying pathophysiology

• Support improved prevention, diagnosis, and treatment of voice disorders

Device and Dataset Overview

The challenge is based on a custom-built smartphone-based system. A lightweight accelerometer is placed just below the Adam’s apple to capture neck-surface vibrations associated with phonation. The sensor connects directly to a smartphone app that records the raw ACC signal and computes clinically meaningful voice measures.

The NeckVibe Challenge dataset includes:

• Patients with PVH and NPVH and age-, sex-, and occupation-matched healthy controls

• Recordings averaging ~11 hours per waking day over about one week per subject

• More than 6,000 hours of ambulatory neck-surface vibration data in total

Further details about feature sets, file formats, and labels are provided on the Tasks & Data page

Significance 

Continuous, real-world tracking of vocal behavior makes it possible to identify maladaptive voice-use patterns, triggers, and contexts that remain hidden in brief lab or clinic assessments. This ecologically valid approach is especially important for VH, where daily voice use plays a central role in onset and persistence.

By transforming neck-surface vibrations into biomarkers of voice function, the NeckVibe Challenge aims to:

• Improve understanding of VH etiology and daily-life impact

• Enable earlier detection and more personalized therapy

• Support real-time biofeedback and remote monitoring

• Extend voice care to individuals in rural or underserved areas who lack access to specialized clinics

Expected Outcomes and Impact

The NeckVibe Challenge is designed to unite researchers, clinicians, and data scientists around a shared goal: advancing interpretable, generalizable models for voice disorder detection based on real-world monitoring.

By participating, teams help:

• Benchmark methods on a large, clinically curated dataset

• Explore new ways to model complex, longitudinal voice behavior

• Move toward scalable, equitable voice health technologies aligned with the Interspeech 2026 theme “Speaking Together”