SDXL 1.0 finetunes on vucinatim/spectrogram-captions for 89 epochs(800 steps). It generates spectrograms for simple sounds. It currently does not produce very good sound effects, but I will train the model for longer in the future.
- Downloads last month
- 4