What if your voice could flag trouble before a scope ever touches your throat? Researchers analyzing a large public voice dataset report that certain acoustic features can distinguish benign vocal fold lesions and laryngeal cancer from healthy voices.
In a new study from Oregon Health and Science University and Portland State University, published in Frontiers in Digital Health, the team examined 12,523 recordings from 306 people in the Bridge2AI-Voice project. They found that the harmonic-to-noise ratio and fundamental frequency, especially in men, carried signals that could help develop a practical voice biomarker for early screening.
What the study did
The team evaluated standardized speech recordings from the initial Bridge2AI-Voice release to see which features best separate laryngeal cancer and benign vocal fold lesions from other voice disorders and from healthy controls. They focused on five familiar acoustic measures: fundamental frequency (F0), jitter, shimmer, mean harmonic-to-noise ratio, and the variability of harmonic-to-noise ratio.
Key findings at a glance
- Across all participants, benign lesions differed from healthy controls in harmonic-to-noise ratio, harmonic-to-noise ratio variability, and fundamental frequency.
- Benign lesions also differed from laryngeal cancer in harmonic-to-noise ratio variability.
- In cisgender men, these patterns were strongest for harmonic-to-noise ratio and its variability.
- No significant differences were detected among cisgender women, likely due to small sample sizes.
Why it matters
Laryngoscopy and biopsy remain the diagnostic gold standard, but they are invasive and often hard to access quickly. A validated voice biomarker could offer a noninvasive prescreen in primary care and telehealth. That could nudge at-risk patients toward faster specialist evaluation. Could a short, standardized reading one day serve as an early warning for laryngeal cancer or track benign vocal fold lesions over time.
What the researchers said
“Here we show that with this dataset we could use vocal biomarkers to distinguish voices from patients with vocal fold lesions from those without such lesions,” said Dr Phillip Jenkins.
Jenkins added, “Our results suggest that ethically sourced, large, multi-institutional datasets like Bridge2AI-Voice could soon help make our voice a practical biomarker for cancer risk in clinical care.”
How the analysis worked
Seven diagnostic cohorts were grouped for two comparisons. Group 1 contrasted laryngeal cancer, benign lesions, and no voice disorder. Group 2 contrasted laryngeal cancer and benign lesions without other voice disorders against spasmodic dysphonia and unilateral vocal fold paralysis. The team used nonparametric methods, including Kruskal–Wallis tests with Dunn’s post hoc comparisons and Holm corrections, then repeated analyses stratified by sex to account for expected differences in pitch and related features.
Important caveats
The authors emphasize that the lack of significant findings among cisgender women may reflect limited sample sizes. Voice features are also influenced by many factors, including age, recording conditions, and coexisting disorders, which means larger, more diverse datasets are essential before clinical deployment.
Where this goes next
“To move from this study to an AI tool that recognizes vocal fold lesions, we would train models using an even larger dataset of voice recordings, labeled by professionals. We then need to test the system to make sure it works equally well for women and men,” said Jenkins. He noted that “Voice-based health tools are already being piloted. Building on our findings, I estimate that with larger datasets and clinical validation, similar tools to detect vocal fold lesions might enter pilot testing in the next couple of years.”
The big picture
Voice biomarker research is moving from theory to practice. This study points to harmonic-to-noise ratio, especially its variability, as a promising feature for distinguishing vocal fold lesions and possibly for early laryngeal cancer detection. With broader sampling and clinical validation, voice analysis could become a low-cost, scalable front door to specialty care.
Journal: Frontiers in Digital Health
Article Title: Voice as a Biomarker: Exploratory Analysis for Benign and Malignant Vocal Fold Lesions
DOI: 10.3389/fdgth.2025.1609811
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.