BINGO aims to understand and decode the neural processes of imagined speech as registered by means of an EEG device. In plain words, BINGO aims to create the essential methodological framework that will translate the imagined pronunciation of words/phonemes/syllables into actual text. BINGO stands over the assumption that EEG signals indeed hold the essential information to discriminate between different imagined speech pronunciations. In order to achieve its ultimate objective (i.e., enabling imagined speech as a robust BCI paradigm), BINGO aims to pass through the following individual challenges that hold the potential to pave the way for natural and intuitive BCI applications.
- Develop, neuro-informed, imagined speech decoding schemes (i.e., algorithms tailored to the neural phenomena of imagined speech) for EEG signals.
- Support an incremental vocabulary under the imagined speech paradigm (i.e., support new words using pre-trained classifiers that can learn incrementally using a small number of new trials).
- Investigate the imagined speech at the level of semantic perception (i.e., compare neural patterns of imagined words of identical meaning expressed in different languages).
- Create a publicly available bench-marking framework (experimental protocol, dataset, evaluation metrics) that enables the fair evaluation of computational models dealing with the aforementioned challenges.