1d csv data or image 2d data for DCGAN

Hi everyone,
i hope that everyone is fine.
i am wondering if you can help me, i have (1600,21,1) signal processing csv data that i made myself(by experimentation ), i am trying hard to train and predict with DCGAN this data but i think that there are lot of thing are missing.
so the question is, do you think that it is better to convert this data into 2d spectogram image and train with it my DCGAN in order to get better performance?

thank you for your answer