Can we apply the model to generate spectrogram extracted from audio ?

[lxrswdd]

So let's say we have spectrogram images of size(256,112) with the number of channels = 1

The data size is very limited, we have about 30 of them. Can we train the DG model on these data? If yes, can you tell us how to do it? Train on customized datasets.

[xhlho]

你好，请问你是否完成了自我数据集的训练，如果已完毕，能麻烦说下怎么实现的吗？