EPG2S: Speech Generation and Speech Enhancement Based on Electropalatography and Audio Signals Using Multimodal Learning | IEEE Journals & Magazine | IEEE Xplore