Audio Generation with Diffusion

Classifier-Guided Inference

Inference method that uses a pre-trained classifier to guide the denoising process toward an output belonging to a specific class (e.g., 'male voice', 'piano'), without modifying the diffusion model weights.

← Terug