https://github.com/Rongjiehuang/FastDiff?tab=readme-ov-file#inference-from-wav-file
I think the essence of fastdiff is to generate wav through mel, so what does it mean to generate wav through wav, since we already have the original wav, why do we need to generate wav？
I know my question may be stupid, it may be because I am a newbie and may not understand fastdiff yet
https://github.com/Rongjiehuang/FastDiff?tab=readme-ov-file#inference-from-wav-file
I think the essence of fastdiff is to generate wav through mel, so what does it mean to generate wav through wav, since we already have the original wav, why do we need to generate wav？
I know my question may be stupid, it may be because I am a newbie and may not understand fastdiff yet