Support LaryngoInsight, a groundbreaking AI research project.
At the heart of this work are Daniel Kumar and Suhaib Ahmed, Indian international students and recent master's graduates of the University of North Texas, working with Dr. Mark Tempesta, Assistant Professor of Voice and Vocal Pedagogy at UNT, voice scientist, and award-winning operatic tenor whose work bridges physics, AI, and the human voice.
Together, they are building an AI system that watches laryngoscopy videos and does what the human eye cannot. It detects subtle patterns in vocal cord movement, frame by frame, and turns them into structured, objective data that researchers can actually use.
The Problem: Voice disorders affect 1 in 3 people at some point in their lives. Yet today, when vocal researchers study how the voice works, they watch laryngoscopy videos by eye. A process that is subjective, inconsistent, and hard to repeat. For musicians especially, where vocal health directly impacts their careers, there is no reliable, objective way to analyze what is happening inside the voice. LaryngoInsight is built to change that.
Why Your Support Is Needed Now
The project has reached a critical turning point, moving from a validated model to a trusted platform for researchers. This next phase requires sustained researcher time and resources to:
- Develop NIH and external grant proposals.
- Strengthen model accuracy and real-world robustness
- Prepare and submit research publications.
- Incorporate ongoing expert feedback into model training
- Refine the annotation interface.
As a proud member of the Indian community, it fills my heart to see Daniel and Suhaib doing work this meaningful on a global stage. Carrying with them the talent, grit, and dedication our community is known for.
What Is LaryngoInsight?
LaryngoInsight is an AI-driven research platform that translates vocal fold video data into objective, structured insights.
By analyzing 33 frame-level parameters including Mode, Density, and Color, it converts complex laryngoscopy patterns into data that researchers and clinicians can actually use.
The project begins with singer health and performance, with a clear path toward speech therapy and broader clinical applications.
The system is trained on expert ratings developed through Complete Vocal Technique (CVT), a globally recognized, science-based vocal framework. Right now it is helping vocal researchers better understand how musicians use and protect their voices, with a clear path toward broader clinical applications in voice health and speech therapy.
How Far They Have Come
This is not just an idea. The system is already live and working.
✅ A full end-to-end pipeline transforms video into structured 33-rating reports
✅ Deep CNN model achieving 83% accuracy.
✅ Advanced multi-backbone model reaching 99.82% internal validation
✅ Expert reviews actively underway.
✅ Continuous model retraining in progress.
What Your Contribution Unlocks
The Bigger Vision - LaryngoInsight is being built for lasting impact across the entire voice health landscape:
- ️A scalable platform for objective voice analysis worldwide.
- Clinical tools for speech therapy and rehabilitation
- Early detection of vocal strain and disorders
Our community has always believed in education, innovation, and lifting each other up. Here is a chance to do all three at once.
Daniel and his team have already done the hard work of proving this is possible. Now they need our community behind them to take it further.
Every contribution, big or small, moves this research one step closer to changing how the world understands the human voice.
Please donate, share, and help spread the word.
For tax deductions or capital gains benefits, business owners and those who itemize may donate directly to the University: Support College of Music
Organizer
