VOWEL SPACE METRICS FOR PEOPLE WITH TYPICAL AND ATYPICAL SPEECH
Johns Hopkins University
Vowel space area (VSA) is an applicable metric for studying speech production deficits and intelligibility. Previous works suggest that the VSA accounts for almost 50% of the intelligibility variance, being an essential component of global intelligibility estimates. However, almost no study publishes a tool to estimate VSA automatically with publicly available codes. On the other hand, neurological disorders (NDs) display a variety of clinical manifestations and represent a challenge to global health. Few diagnostic metrics can be reliably employed to easily distinguish NDs without costly imaging techniques or invasive procedures. Vowel articulation from speech can be assessed in a non-invasive manner and can provide biomarkers of NDs. In this study, we proposed an open-source tool called VSAmeter to measure VSA and vowel articulation index (VAI) automatically and validate it with the VSA and VAI obtained from a dataset in which the formants and phone segments have been annotated manually. The results show that VSA and VAI values obtained by our proposed method strongly correlate with those generated by manually extracted F1 and F2 and alignments. Such a method can be utilized in speech applications, e.g., the automatic measurement of VSA and VAI for the evaluation of speakers with dysarthria. We employed VSAmeter to evaluate the articulation of vowels from people with different NDs: Parkinson’s and Alzheimer’s Disease, and Parkinson’s mimics. The effect of different types of speech tasks on distinguishing people with different NDs was considered in this study. In addition to VSA and VAI, a new metric, Cumulative Pair-wise Vowel Distance (CPVD), was explored, measuring the top k shortest distance pairs between vowels in the F1-F2 space. The new metric provided significant differences between Parkinson’s and Parkinson’s mimics, and the control group in two speech tasks.
Vowel space area (VSA), Vowel articulation index (VAI), Automatic assessment, Correlation analysis, Speech and language technologies, Neurological Disorders, Atypical Speech, Parkinson’s disease