Y. Liang, “SLD-L2S: Hierarchical Subspace Latent Diffusion for High-Fidelity Lip to Speech Synthesis”, AAAI, vol. 40, no. 38, pp. 31943–31951, Mar. 2026.