CapSpeech is a fully open-source project for academic use, offering a large-scale pretraining dataset, high-quality fine-tuning subsets, & pretrained models.

→ View original post on X — @mit_csail