Common Spatial Pattern Based Data Augmentation Technique for Decoding Imagined Speech