@misc{10481/105970,
year = {2025},
month = {8},
url = {https://hdl.handle.net/10481/105970},
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.},
organization = {This work was supported by the grant PID2022-141378OB-C22 funded by MICIU/AEI/10.13039/501100011033 and ERDF/EU.},
publisher = {ISCA},
title = {Recreating Neural Activity During Speech Production with Language and Speech Model Embeddings},
doi = {10.21437/Interspeech.2025-1400},
author = {Khanday, Owais Mujtaba and Rodríguez San Esteban, Pablo and Ahmad, Zubair and Ouellet, Marc and González López, José Andrés},
}