We've published a paper in Nature Digital Medicine on the feasilbility of computer assisted voice analysis for Ataxias. Ataxias are a group of movement disorders that are characterized by progressive loss of balance, impaired coordination and speech disturbance, which together lead to markedly reduced quality of life. Speech disturbance is clinically diagnosed, but methods for objective assessment of severity are lacking. Using 71 sets of speech recordings from ataxia patients, we developed an automated classification system. With a tolerance of ±1 point, this classification system correctly predicted experts’ ratings of speech disturbance according to item 4 of the Scale for Assessment and rating of ataxia (SARA) in 80% of cases. We thereby demonstrate feasibility of computer-assisted voice analysis for automated assessment of severity of speech disturbance.
PeakProfiling
Pioneers in advanced sound analytics
You have lots of voice data but you are unsure if / how you can process them in a compliant way with optimal technology.
You want to measure the emotiol state of a speaker from their voice and need a cutting-edge technology partner.
You tried to build a voice biomarker on your own, but success rates (accuary, sensitivities, specificities, AUCs, etc.) were too low.
You want to predict a critical type of defect in your machine, but conventional approaches at predictive maintenance are unsuccessful.
You want a specific lifetime prediction or real-time information on the remaining lifetime of your machine/ component, not a general estimate by the manufacturer or subjective human judgement.
You feel like Industry 4.0 is a buzzword and you want to see a hands-on solution.