Comparison of In-Person and Online Recordings in the Clinical Teleassessment of Speech Production: A Pilot Study

Abstract

In certain circumstances, speech and language therapy is proposed in telepractice as a practical alternative to in-person services. However, little is known about the minimum quality requirements of recordings in the teleassessment of motor speech disorders (MSD) utilizing validated tools. The aim here is to examine the comparability of offline analyses based on speech samples acquired from three sources: (1) in-person recordings with high quality material, serving as the baseline/gold standard; (2) in-person recordings with standard equipment; (3) online recordings from videoconferencing. Speech samples were recorded simultaneously from these three sources in fifteen neurotypical speakers performing a screening battery of MSD and analyzed by three speech and language therapists. Intersource and interrater agreements were estimated with intraclass correlation coefficients on seventeen perceptual and acoustic parameters. While the interrater agreement was excellent for most speech parameters, especially on high quality in-person recordings, it decreased in online recordings. The intersource agreement was excellent for speech rate and mean fundamental frequency measures when comparing high quality in-person recordings to the other conditions. The intersource agreement was poor for voice parameters, but also for perceptual measures of intelligibility and articulation. Clinicians who plan to teleassess MSD should adapt their recording setting to the parameters they want to reliably interpret.

Keywords: apraxia of speech; dysarthria; dysphonia; motor speech disorders; recording devices; speech analysis; speech and language therapy; speech and voice disorders; teleassessment; telerehabilitation.

Figure 1

Setting for the simultaneous recording from three sources, namely (1) high-quality in-person hardware (local HQ condition), (2) in-person standard hardware, with software automatically sending audio files online (local SQ condition), and (3) remote standard hardware (online condition).