train AudioLDM from beginning to end

[Linghuxc]

@haoheliu Hi haohe, I'm very interested in your work. It's a great piece of work. I want to train audioldm from scratch. Should I train a VAE first and then the latent diffusion part?
Does that mean that the trained VAE is used in the process of training latent diffusion and the parameters of the VAE model are stored in the model parameters of the ldm?
Perhaps I would like to know a complete training process, thank you very much!