v1: ONNX bundle with voices, PyTorch weights.
Note ONNX bundle with the voice library.
Note PyTorch weights for training, finetuning and voice export.
Note Shared speech codec (autoencoder).