Speech disability or verbal communication loss caused by injuries and neurodegenerative diseases that affect the brain, speech articulation, and language understanding affects over 2.2% of world population. 

Currently, a new brain computer interfacing (BCI) is under research to offer a bridge between the brain waves and outer world. In such thought-to-speech interfaces people who cannot speak can use their brain signals to communicate. 

To extend the current EEG-to-speech methods to error-free natural speech production from imagined speech, or rehabilitation of human speech production system, another step forward is needed to secure both error-free and semantic speech production specially for limited data and computational resources. 

Therefore, for the first time, in this project we will attempt visual feedback together with estimation of emotion type and level from the brain by means of electroencephalography (EEG) recordings. In this scenario, a visual system shows the most-probable responses of EEG-to-speech translation system and through visual feedback, the subject selects the correct response to be spoken. In addition, the emotion type and level are estimated from the EEG and imposed on the produced speech making it natural. The emotion information will be used to change the morphology of generated speech to express the emotion of speakers in their speech. As the result, we will have a real-time EEG-to-speech prediction and generation system which can be used for two major applications; one for generation of imagined speech for those with complete speech disability and second to rehabilitate people suffering from partial speech disability such as those suffering from autism spectrum disorder, stroke, or depression.  

A large community with complete or partial speech disabilities, can benefit from the project outcome.