We use high-density electrocorticographic (HD-ECoG) recordings from the human sensorimotor cortex to study the neural basis of speech production and to develop computational models for speech decoding. Using deep learning techniques, we aim to decode not only spoken words but also the underlying articulatory movements that generate speech. By restoring articulatory movement trajectories from neural activity, this work seeks to enable more naturalistic and flexible speech synthesis. Ultimately, this research contributes to the development of brain–computer interfaces that can restore communication in individuals who have lost the ability to speak due to injury or progressive neurological disease.