16th Annual Conference of the International Speech Communication Association

Dresden, Germany
September 6-10, 2015

Temporal Dynamics of the Speech Readiness Potential, and Its Use in a Neural Decoder of Speech-Motor Intention

Jonathan S. Brumberg, Nichol Castro, Akshatha Rao

University of Kansas, USA

We investigated the temporal dynamics of brain activity related to speech and motor preparation. Previous electroencephalography (EEG) studies of speech production have identified a slow wave negativity that occurs as early as 2 seconds prior to articulation, known as the readiness potential (RP). This EEG potential is commonly described as having two phases, an early slow component and a late fast component. In our study, we collected RPs from 15 subjects and fit a linear spline with four fixed knots and their locations as free parameters. The result was a piecewise linear approximation to the subject RPs from which it was possible to identify the onset and termination of both the slow and fast components, as well as their slopes. In addition, we used the collected RPs to train an adaptive filter neural decoding algorithm to predict occurrences of RPs from trial-based epochs of imagined speech and motor movements for use in a brain-computer interface for speech communication. The initial spline analysis will help to determine the contribution of each RP phase to the representations of intended speech-motor behavior, reducing the complexity of the adaptive filter for more efficient use in real-time.

Full Paper

Bibliographic reference.  Brumberg, Jonathan S. / Castro, Nichol / Rao, Akshatha (2015): "Temporal dynamics of the speech readiness potential, and its use in a neural decoder of speech-motor intention", In INTERSPEECH-2015, 1126-1130.