Figure 5
From: Decoding of the speech envelope from EEG using the VLAAI deep neural network
Evaluation of the baseline models and the VLAAI network on the single-speaker trials (50 seconds per trial) of the DTU dataset. While the reconstruction scores are lower for all models compared to the single-speaker stories dataset (see also Fig. 2), the general conclusions still hold. The VLAAI network significantly outperforms all other baseline models (p<0.01), with a relative performance increase of 62% over the linear decoder. (n.s.: p \(\le\) 0.05, *: 0.01 \(\le\) p < 0.05, **: 0.001 \(\le\) p < 0.01, ***: p < 0.001)
