We developed dysarthric speech intelligibility classifiers on 551,176 disordered speech samples contributed by a diverse set of 468 speakers, with a range of self-reported speaking disorders and rated for their overall intelligibility on a five-point scale. We trained three models following different deep learning approaches and evaluated them on ~94K utterances from 100 speakers. We further found the models to generalize well (without further training) on the TORGO database (100% accuracy), UASpeech (0.93 correlation), ALS-TDI PMP (0.81 AUC) datasets as well as on a dataset of realistic unprompted speech we gathered (106 dysarthric and 76 control speakers,~2300 samples).
@article{arxiv.2303.07533,
title = {Speech Intelligibility Classifiers from 550k Disordered Speech Samples},
author = {Subhashini Venugopalan and Jimmy Tobin and Samuel J. Yang and Katie Seaver and Richard J. N. Cave and Pan-Pan Jiang and Neil Zeghidour and Rus Heywood and Jordan Green and Michael P. Brenner},
journal= {arXiv preprint arXiv:2303.07533},
year = {2023}
}