With the increased presence of wearables, photoplethysmography (PPG) is a promising alternative to electrocardiography (ECG) for the monitoring of physiological parameters related to cardiovascular diseases. However, the evaluation of PPG-based methods is more challenging since PPG signals are less frequently available in databases compared to ECG signals. Generative adversarial networks (GANs) are a promising alternative to synthesize PPG signals and compensate for unavailable PPG recordings. We propose a modified GAN architecture to generate PPG signals from a sequence of spikes located at R-peak positions of the ECG. The validity of the synthetic PPG signals was evaluated in two ways. First, by comparing the statistical distribution of the pulse wave morphology features of real vs. synthetic PPGs. Second, by comparing the interbeat-intervals (IBIs) of the synthetic PPGs vs. the ECG-based reference IBIs. The statistical distributions did not reveal significant difference between real and synthetic PPGs. In Addition, the IBIs extracted from synthetic PPGs were accurately reproducing the ECG-based reference IBIs, with a mean absolute error of 10.24 ms. Overall, the results highlight the potential of synthetic PPG to optimize and validate PPG-based method for cardiovascular monitoring.