Hidenori Sugano

This paper synthesizes speech from imagined speech ECoG using a Transformer decoder and pre-trained vocoder, achieving high correlation with proxy targets.

This paper demonstrates that closed-loop brain-computer interfaces can retrieve intended images using a CLIP latent space, achieving over 80% accuracy, while failing with AlexNet.

This study uses a Transformer model trained on overt speech to decode covert speech from ECoG data, achieving similar performance to models trained directly on covert speech.

This paper introduces data-driven neural networks with linear-phase filters for detecting epileptic spikes in EEG data, achieving an average AUC of 0.967.