Recreating Neural Activity During Speech Production with Language and Speech Model Embeddings

Abstract

Understanding how neural activity encodes speech and language production is a fundamental challenge in neuroscience and artificial intelligence. This study investigates whether embeddings from large-scale, self-supervised language and speech models can effectively reconstruct high-gamma neural activity characteristics, key indicators of cortical processing, recorded during speech production. We use pre-trained embeddings from deep learning models on linguistic and acoustic data to map high-level speech features onto high-gamma signals. We analyze the extent to which these embeddings preserve the spatio-temporal dynamics of brain activity. Reconstructed neural signals are evaluated against high-gamma ground-truth activity using correlation metrics and signal reconstruction quality assessments. The results indicate High-gamma activity was effectively reconstructed using language and speech model embeddings, yielding Pearson correlation coefficients of 0.79–0.99 across all participants.