Intelligibility of modifications to dysarthric speech

Abstract

Dysarthria is a motor speech impairment affecting millions of people. Dysarthric speech can he far less intelligible than that of non-dysarthric speakers, causing significant communication difficulties. The goal of this work is to understand the effect that certain modifications have on the intelligibility of dysarthric speech. These modifications are designed to identify aspects of the speech signal or signal processing that may he especially relevant to the effectiveness of a system that transforms dysarthric speech to improve its intelligibility. A result of this study is that dysarthric speech can, in the hest case, he modified only at the short-term spectral level to improve intelligibility from 68% to 87%. A baseline transformation system using standard technology, however, does not show improvement in intelligibility. Prosody also has a significant @ < 0.05) effect on intelligibility.

Articulatory Knowledge in the Recognition of Dysarthric Speech

“…Since each speaker in the TORGO database recites the same set of phrases, we achieve frame-by-frame alignment by applying dynamic time warping on corresponding acoustic segments of preannotated speech, and applying the resulting alignment on the raw articulatory data. This is effectively the reverse of the approach suggested by Hosom et al, who propose transforming dysarthric acoustic space to regular acoustic space in order to be made more intelligible [69].…”

Intelligibility Assessment and Speech Recognizer Word Accuracy Rate Prediction for Dysarthric Speakers in a Factor Analysis Subspace

“…The iVector of utterance s was calculated as the expectation of the posterior distribution of i given the sufficient statistics of the utterance [Dehak et al 2011]. The main difference between the common formulation of FA [Bishop 2006] and ours is that in the former, the latent variable changes for every frame, whereas in our case, the latent variable is common to the whole utterance [Kenny et al 2007]. …”

Phonological features in discriminative classification of dysarthric speech

“…These disorders typically limit motor function generally, making other physical interaction (e.g., keyboard) slower, and less desirable than spoken expression [1]. Unfortunately, automatic speech recognition is currently ill-suited to dysarthric speech, rendering such software inaccessible to those who might most benefit from it.…”