a text-conditional diffusion probabilistic model capable of generating high fidelity audio. - View it on GitHub