Assessing Neuromotor Coordination in Depression Using Inverted Vocal Tract Variables

Carol Espy-Wilson, Adam C. Lammert, Nadee Seneviratne, Thomas F. Quatieri


Speech articulation is a complex activity that requires finely timed coordination across articulators, i.e., tongue, jaw, lips, and velum. In a depressed state involving psychomotor retardation, this coordination changes and in turn modifies the perceived speech signal. In previous work, we used the correlation structure of formant trajectories as a proxy for articulatory coordination, from which features were derived for predicting the degree of depression. Ideally, however, we seek coordination of the actual articulators using characteristics such as the degree and place of tongue constriction, often referred to as a tract variable (TV). In this paper, applying a novel articulatory inversion process, we investigate the relation between correlation structure of formant tracks versus that of TVs. We show on a pilot depressed/control dataset that, with the same number of variables, TV coordination-based features, although with some characteristics similar to their counterpart, outperform the corresponding formant track correlation features in detection of the depressed state. We speculate on the latent information being captured by TVs that is not present in formants.


 DOI: 10.21437/Interspeech.2019-1815

Cite as: Espy-Wilson, C., Lammert, A.C., Seneviratne, N., Quatieri, T.F. (2019) Assessing Neuromotor Coordination in Depression Using Inverted Vocal Tract Variables. Proc. Interspeech 2019, 1448-1452, DOI: 10.21437/Interspeech.2019-1815.


@inproceedings{Espy-Wilson2019,
  author={Carol Espy-Wilson and Adam C. Lammert and Nadee Seneviratne and Thomas F. Quatieri},
  title={{Assessing Neuromotor Coordination in Depression Using Inverted Vocal Tract Variables}},
  year=2019,
  booktitle={Proc. Interspeech 2019},
  pages={1448--1452},
  doi={10.21437/Interspeech.2019-1815},
  url={http://dx.doi.org/10.21437/Interspeech.2019-1815}
}