Crowdsourcing may help to assess voice training in Parkinson’s: Study

Approach found to accurately capture changes in patients' voice quality

Marisa Wexler, MS avatar

by Marisa Wexler, MS |

Share this article:

Share article via email
A person uses a megaphone to make an announcement.

Crowdsourcing — collecting data from large groups of individuals via the internet — could be used in clinical studies to assess the real-world effect of voice and speech therapy on voice quality in people with Parkinson’s disease, according to a recent study.

Researchers used crowdsourcing to evaluate speech samples of Parkinson’s patients before and after such training, “finding a significant change in perceptually rated voice quality from pre- to posttreatment that remained significant at the 6-month follow-up time point.”

“The findings reported here support the validity of crowdsourced listeners’ ratings of voice quality in a broad sense,” the researchers wrote.

The paper, “Crowdsourced Perceptual Ratings of Voice Quality in People With Parkinson’s Disease Before and After Intensive Voice and Articulation Therapies: Secondary Outcome of a Randomized Controlled Trial,” was published in the Journal of Speech, Language, and Hearing Research.

Recommended Reading
A doctor and patient use a computer for a telehealth visit.

Parkinson Voice Project Seeking $20M to Expand SPEAK OUT! in US

Voice quality problems common in Parkinson’s

People with Parkinson’s disease often experience issues with speech, such as an unusually quiet voice or difficulty enunciating words. Speech therapy such as the Lee Silverman Voice Treatment — a program of vocal exercises developed in the 1980s and known as LSVT LOUD — may help improve patients’ ability to make themselves understood.

Prior research has shown that interventions such as LSVT LOUD can lead to clinical improvements in voice quality for Parkinson’s patients. To date, studies have generally relied on either highly trained experts or computer-based analyses to measure voice quality.

But there is a dearth of data on how such training impacts patients in their day-to-day lives.

“Studies of speech and voice treatment methods for individuals with Parkinson’s disease typically evaluate progress using fine-grained acoustic measures or highly trained listeners’ perceptual ratings,” Tara McAllister, PhD, a professor at New York University and co-author of the study, said in a press release.

“However, this approach leaves some uncertainty about the real-world impact of these changes, since it is not always clear they would be apparent to untrained listeners that patients interact with in their daily life,” McAllister said.

While crowdsourcing was originally developed for commercial purposes (e.g., to verify the accuracy of online product listings), it has been adopted by behavioral researchers as a convenient means to gain access to a large number of individuals considered representative of the general population.

Now, researchers sought to test whether crowdsourcing could be used to reliably assess voice quality in a way that might be more reflective of real-world experiences. Crowdsourcing involves large online platforms where individuals can sign up to do tasks, oftentimes in exchange for monetary compensation.

“While crowdsourcing was originally developed for commercial purposes (e.g., to verify the accuracy of online product listings), it has been adopted by behavioral researchers as a convenient means to gain access to a large number of individuals considered representative of the general population,” the researchers wrote.

Recommended Reading
Doctors, scientists, and other experts talk with Expert Voices written in front of them.

Expert Voices: Dealing With Parkinson’s Disease and Speech Difficulties

Voice quality evaluations similar for crowdfunding, expert raters

For this study, the researchers used data from an earlier clinical trial that evaluated the LSVT LOUD intervention — a total of 250 vocal recordings from 85 people, 65 with Parkinson’s and 20 without, taken before and after the intervention, as well as six months after treatment. The results of the original study had suggested that this intervention led to significant improvements in the Acoustic Voice Quality Index (AVQI), a computer-based measure of vocal quality.

“This study also extends the findings of [the earlier clinical trial] by testing whether the changes in voice quality that they measured acoustically would be perceptually detected by everyday listeners,” the scientists wrote.

Using a crowdfunding platform, 29 people evaluated these recordings. For each recording, the participant marked whether the voice sounded typical or atypical; this binary division was chosen to minimize confusion on the part of the raters.

Analyses of the results showed that the individual crowdfunding participants generally showed consistency over time in how they rated voices, and ratings were generally similar across the raters. Further analyses showed a significant correlation between AVQI scores and the number of participants who rated samples as atypical, and in line with the original study, data suggested an improvement in vocal quality for patients given the LSVT LOUD intervention.

“The crowdsourced ratings reported here indicate that LSVT LOUD participants’ change in voice quality was perceptually apparent to listeners with no previous training related to speech or voice, supporting the real-world relevance of the findings reported by” the original clinical trial, the scientists concluded.

More broadly, these findings support the idea that crowdsourcing could be used as a way to measure the real-world impact of interventions on voice quality in Parkinson’s patients. The researchers noted, however, that further work is needed to validate these findings and determine the most helpful ways to use crowdsourcing in clinical research.