Encoding of speech modes and loudness in ventral precentral gyrus

Area of Science:

• Neuroscience
• Speech Communication
• Biomedical Engineering

Background:

• Human vocal communication relies on varying speech mode and loudness for expression.
• The neural encoding of speech mode and loudness in the ventral precentral gyrus (vPCG) at the neuronal firing rate level remains largely uninvestigated.

Purpose of the Study:

• To investigate the encoding of speech mode and loudness in the vPCG using high-resolution neuronal firing rates.
• To assess the feasibility of decoding attempted speech mode and loudness from vPCG activity.
• To develop and evaluate a closed-loop system for modulating speech neuroprosthesis output based on attempted loudness.

Main Methods:

• Utilized intracortical microelectrode arrays implanted in the vPCG of two participants in a speech neuroprosthesis clinical trial.
• Recorded neuronal firing rates during attempted mimed, whispered, normal, and loud speech.
• Applied decoding algorithms to analyze neural ensemble activity and identify distinct neural subspaces for mode/loudness and phonemic content.

Main Results:

• Neuronal firing rates in the vPCG showed strong modulation based on attempted speech mode and loudness.
• Attempted mode/loudness was decoded with high accuracy (94% and 89%) from vPCG activity.
• Neural preparatory activity preceding speech onset allowed for significant decoding accuracy (80%).
• A closed-loop loudness decoder achieved 94% online accuracy in modulating a brain-to-text speech neuroprosthesis.

Conclusions:

• Demonstrated the feasibility of decoding speech mode and loudness from the vPCG.
• Findings support the potential for developing speech neuroprostheses capable of producing more expressive vocalizations.
• Highlights the role of vPCG in encoding dynamic aspects of speech production.