Improving EEG based Continuous Speech Recognition

Improving EEG based Continuous Speech Recognition Gautam Krishna Brain Machine Interface Lab The University of T exas at Austin Austin, Texas Co Tran Brain Machine Interface Lab The University of T exas at Austin Austin, Texas Mason Carnahan Brain Machine Interface Lab The University of T exas at Austin Austin, Texas Y an Han Brain Machine Interface Lab The University of T exas at Austin Austin, Texas Ahmed H Tewfik Brain Machine Interface Lab The University of T exas at Austin Austin, Texas Abstract --In this paper we introduce various techniques to improve the performance of electroencephalography (EEG) features based continuous speech recognition (CSR) systems. A connectionist temporal classification (CTC) based automatic speech recognition (ASR) system was implemented for performing recognition. We introduce techniques to initialize the weights of the recurrent layers in the encoder of the CTC model with more meaningful weights rather than with random weights and we make use of an external language model to improve the beam search during decoding time. We finally study the problem of predicting articulatory features from EEG features in this paper . ASR systems forms front end or back end in many state of the art voice assistant systems like Bixby, Alexa,Siri,Cortana etc.