Toward EEG Sensing of Imagined Speech

Might EEG measured while one imagines words or sentences provide enough information for one to identify what is being thought? Analysis of EEG data from an experiment in which two syllables are spoken in imagination in one of three rhythms shows that information is present in EEG alpha, beta and theta bands. Envelopes are used to compute filters matched to a particular experimental condition; the filters' action on data from a particular trial lets one determine the experimental condition used for that trial with appreciably greater-than-chance performance. Informative spectral features within bands lead us to current work with EEG spectrograms.

• Type: Journal
• Year: 2009
• Where: HCI
• Authors: Michael D'Zmura, Siyi Deng, Tom Lappas, Samuel Thorpe, Ramesh Srinivasan
• Keywords: EEG Alpha, EEG Data, EEG Spectrograms, HCI 2009, Human Computer Interaction