I have a set of noisy real data that I have no any information of the noise that is corrupted by it. I denoised it but I do not know how to show enhancement.
Normally to calculate SNR you should have clear signal (signal of interest) and noise signal separated, then you can calculate signal power and noise power. In your case, you have only the mix of this two components. Therefore maybe you can find in this signal some fragments (in manual or automatic way) where the signal does not exist thus remaining part is only noise (i.e. in speech signal there are some moments when nobody speaks) then you can cut these fragments and next calculate the average power of noise. In the same way you can find fragments where there exist strong signal. Based on the assumption that SNR is high you can calculate the power of the signal of interest. Then you can calculate the real SNR approximate. The same procedure can be used for signal before and after denoising.