Sensing Your Vocals: Exploring the Activity of Vocal Cord Muscles for Pitch Assessment Using Electromyography and Ultrasonography

Paper Title

Sensing Your Vocals: Exploring the Activity of Vocal Cord Muscles for Pitch Assessment Using Electromyography and Ultrasonography

Publication Info

• Topic area: Vocal training enhancement using physiological sensing technologies.
• Keywords: Vocal training, Electromyography (EMG), Ultrasonography (UI), Vocal cord muscles, Pitch assessment, Singing pedagogy, Physiological sensing, Human-Computer Interaction (HCI), Singing Power Ratio (SPR), Vocal performance.

Background and Problem

• Problem / challenge: Traditional vocal training methods rely on auditory feedback and spectrograms, which fail to provide direct insights into the muscle mechanisms controlling pitch. This leaves singers to rely on trial and error to achieve desired vocal outcomes.
• Significance: Understanding and visualizing vocal muscle activity can improve vocal training by offering more intuitive and precise feedback, addressing the limitations of current methods.
• Motivation and related work: Prior research has explored metrics like the Singing Power Ratio (SPR) and respiratory muscle training but lacks intuitive connections between sound and muscle activity. Emerging sensing technologies like EMG and UI offer potential for addressing these gaps but remain underexplored in vocal pedagogy.

Solution

• Proposed approach: Leveraging Electromyography (EMG) and Ultrasonography (UI) to visualize and assess vocal cord muscle activity for improving vocal training.
• Novelty:
  1. Creation of a Vocal Cord Sensing Dataset (VCSD) using EMG and UI from singers of varying skill levels.
  2. Development and evaluation of feedback systems for novice and experienced singers, comparing EMG and UI with traditional methods.
  3. Recommendations for integrating sensing technologies into vocal training systems, with insights into their broader applications.
• Procedure and key techniques:
  1. Study 1: Collected and analyzed EMG and UI data from 16 singers to establish baseline patterns of vocal muscle activity and differentiate skill levels.
  2. Study 2: Evaluated the impact of EMG and UI feedback on 12 novice singers, focusing on skill development and usability.
  3. Study 3: Investigated the performance of a portable EMG setup with 15 experienced singers, comparing it to traditional feedback methods.

Results

• Concrete findings:
  • EMG effectively captured temporal stability of vocal muscles, while UI provided detailed visualizations of vocal cord dynamics.
  • Novices showed improved vocal control with both methods, though EMG enhanced perceived controllability and UI improved vocal cord length control.
  • Experienced singers demonstrated strong coordination between EMG and SPR, with tenors showing more consistent muscle engagement than bass singers.
• Advantage over baselines:
  • EMG and UI provided more precise and actionable feedback than traditional auditory methods.
  • EMG reduced vocal fatigue and enhanced real-time control, while UI offered intuitive visual cues for muscle activity.
• Experiments / evaluation:
  • Study 1: 16 participants; analyzed EMG and UI data across skill levels.
  • Study 2: 12 novices; compared audio-only, EMG, and UI feedback using NASA-TLX and SoAS metrics.
  • Study 3: 15 experienced singers; examined EMG-SPR correlations and conducted focus groups.
• Limitations and future work:
  • Short-term training effects; need for longer-term studies.
  • Cognitive load in interpreting feedback; requires more intuitive interfaces.
  • Gender and developmental confounds in expert baseline comparisons.
  • Small sample size and variability in group performance.

Summary

This paper demonstrates the potential of EMG and UI for enhancing vocal training by providing detailed feedback on vocal muscle activity. EMG excels in capturing temporal stability and improving perceived control, while UI offers intuitive visualizations of vocal cord dynamics. Through three studies involving 43 participants, the authors show that these methods outperform traditional auditory feedback, with applications extending to speech therapy and music education. Future work will focus on refining interfaces, addressing cognitive load, and exploring long-term training effects.