Intelligibility and acoustic analysis of speech production in synthetic and natural reverberant and noisy environments: A case study of Parkinson’s disease speech

Abstract

Our understanding of the effects of noise and reverberation on speech production and intelligibility is enhanced by the study of speech from talkers with Parkinson’s disease (PD) due to the fact that PD speech spans a wide range of intelligibility in quiet and that talkers with PD also span a wide range of speech motor control for compensation. A previous study on the intelligibility of Parkinson’s disease (PD) speech in a synthetic reverberant and noisy environment showed that speech intelligibility suffered primarily due to noise although the effects of reverberation were also significant [Shrivastav, ASHA 2011]. A follow-up study in natural reverberant and noisy environments (allowing for talker compensation) found that the decrease in speech intelligibility as speaking conditions worsened was proportional to dysarthria severity [Kopf et al., ASHA 2013], and an acoustic analysis quantified the range of compensations made by talkers with PD [Shrivastav et al., ASA May 2014]. The current study modeled the intelligibility data from the synthetic environment study and applied the model to the data from the natural environment study to quantify the effects of talker compensation on intelligibility and to relate acoustic measures of speech to talker compensation.