Data augmentation for audio is a pivotal technique that involves creating additional training samples by applying transformations like pitch shifting or time stretching to existing audio clips. This is especially important when working with small data sets, as it greatly increases the performance of deep learning models. Advanced methods, such as WAVEGAN, a type of Generative Adversarial Network (GAN), are used to further enhance the dataset. WAVEGAN is specially adapted for the uncontrolled synthesis of sound with a raw curve and its application allows the generation of realistic sound samples. These synthesized samples find versatile applications, from speech synthesis to enhancing sound effects in media, and even training speech recognition systems. By harnessing WAVEGAN, the training dataset can be efficiently expanded, ultimately leading to more precise audio classification.