CONFERENCE / ICCAIS-2026

In recent years, Generative Artificial Intelligence has shown remarkable progress in creating realistic multimedia content. One interesting application of this technology is talking face generation, where a still image of a person can be animated according to a given speech signal. The main challenge in such systems is to ensure that the lip movements correctly match the spoken words. If synchronization is not accurate, the generated video appears unnatural. In this work, we present a lip synchronization system based on Generative Adversarial Networks (GANs). The proposed model takes a static facial image and a speech audio file as inputs and produces a talking video in which the lip movements are aligned with the audio. The adversarial learning strategy helps improve both synchronization accuracy and visual quality of the generated frames.