Geometry of orofacial neuromuscular signals: speech articulation decoding using surface electromyography

Harshavardhana T Gowda, Zachary D. McNaughton, Lee M. Miller

Abstract:

*Objective.*
In this article, we present data and methods for decoding speech articulations using surface electromyogram (EMG) signals. EMG-based speech neuroprostheses are useful for restoring audible speech in individuals who have lost the ability to speak intelligibly due to laryngectomy, neuromuscular diseases, stroke, or trauma-induced damage (e.g., radiotherapy toxicity) to speech articulators.

*Approach.*
To achieve this, we collect EMG signals from the face, jaw, and neck as subjects perform orofacial movements or articulate speech. These signals are recorded using surface electrodes placed at key muscle sites involved in speech production. Furthermore, we design and evaluate models that efficiently decode EMG with limited data.

*Main results.*
Our findings reveal that the manifold of symmetric positive definite (SPD) matrices serves as a natural embedding space for EMG signals. Specifically, we provide an algebraic interpretation of the manifold-valued EMG data using linear transformations and analyze and quantify EMG signal distribution shifts across individuals.

*Significance.*
Overall, our approach demonstrates significant potential for developing parameter-efficient neural networks that can be trained with minimal data. This is particularly important for embedded sensor devices with constrained computational resources. Additionally, given the challenges associated with collecting EMG data at scale, our method potentially offers a practical solution for advancing noninvasive speech neuroprostheses.