Correlation of acoustic features of pitch/rhythm/power and perceptual impressions after singing training for people with dysarthria

The aim of this fundamental study is to investigate acoustic changes and perceptual impressions by singing and vocal training in 11 dysarthric Japanese patients. We also examined to improve their speech intelligibility through this training. Dysarthria is a speech disorder caused by diseases such as stroke, intractable neurological disease, and sequelae of head trauma. In terms of vocal evaluations, pre-test, mid-test, post-test, and follow-up test were performed quarterly to investigate a correlation between the five acoustic features of the singing voice (i.e. normalized frequency (pitch) score (NFS), normalized rhythm (duration) score (NRS), normalized intensity (power) score (NIS), frequency deviation (FD) and the intensity deviation (ID)) and speech intelligibility. These factors were evaluated based on the perceptual impression of a group of music major students (MS) and a group of non-music major students (NMS). The objective acoustic features revealed that the order of NRS, NIS, NFS, FD, and ID was higher in correlation with human subjective evaluation. The pre-test results of speech intelligibility indicated a greater improvement in the low intelligibility group than in the high intelligibility group. Furthermore, no difference was found in how perceptual impressions were evaluated between the MS and NMS groups.