Spaces:

artificialguybr

/

Stable-Audio-Open-Zero

like 330

Running on Zero

FYI (because a lot of the example prompts here seem to be using this style), you don't need to tell this model to do things instruct-style. "Generate a...", "Synthesize a...", "Create me a..." will all likely just degrade the output quality. The model isn't trained with prompt-rewriting, so it's learned from the raw text descriptions and metadata on Freesound. Just describe the sound you want rather than talking to the model like ChatGPT, and you'll get better results :)