[pledess]

Another possibility is ML for Cued Speech: https://arxiv.org/abs/2212.01083