Audio Super Resolution with AudioSR. It estimates a dynamic gpu duration done by a private Benchmarking HuggingFace ZeroGPU (H200) Space on the 16th November 2025 for saving quota.
Guidance scale (Large => better quality and relavancy to text; Small => better diversity)
The sampling step for DDIM
Changing this value (any integer number) will lead to a different generation result, put 0 for a random one.