Fingerprints are regarded as the most reliable form of human identification for thousands of years. Even though the fingerprint acquisiton process has became more convenient with technological advancements, privacy concerns hindered the data collection and, hence advancement of research on fingerprint biometrics. Like many other problem solved with deep learning, biometrics also requires a sizable database to succeed. This study focuses on synthetic fingerprint generation to tackle bottlenecks created by data scarcity. First, a preprocessing pipeline is designed tp enhance images from a small publicly available fingerprint dataset. Next the new enhanced dataset is given as an input to a generative network to create candidate synthetic fingerprints. Lastly Fingerprint image quality models filter low-quality fingerprint images from the candidate set to form the synthetic fingerprint dataset. Numerous experiments were conducted to show the usability of the generated synthetics fingerprints using both real and synthetic fingerprint datasets available for network trainig. Experimental results show that enhancing fingerprint images from real-life datasets helps models trained with synthetic fingerprint images classify enhanced versions of the real-life fingerprint samples.Synthetic fingerprints generated using the proposed pipeline can establish a good training set which can imporove deep neural network performance as substantially as their real- life counterparts, but without introducing any privacy concerns.