Brain-to-speech decoder: speech envelope decoding

Research project that compares models for reconstructing low-frequency amplitude envelopes from English listeners' electroencephalographic (EEG) signals while listening to an English audiobook. Data are from Reetzke, Gnanateja, and Chandrasekaran (2021). This research compares the envelope decoding performance (as measured by Pearson correlation) of the multivariate temporal response function (mTRF) and an LSTM-based deep neural network (DNN) model. This reposoitory contains the code for the DNN model.