Voltage stability is a critical challenge in modern powersystems, especially in developing nations such as Nigeria, where aging infrastructure, high load demand, and intermittent renewable integration threaten system security. Traditional methods for stability assessment rely on linearized models, sensitivity indices, or contingency simulations, which often fail to capture nonlinear patterns under stressed grid conditions. This paper presents a deep learning–based offline voltage stability monitoring framework for the Yauri 330 kV Transmission Company of Nigeria (TCN) substation. A hybrid Long Short-Term Memory (LSTM) and Conditional Generative Adversarial Network (cGAN) is adopted to overcome data scarcity and improve the robustness of prediction. The framework incorporates theoretical formulations of LSTM cell dynamics, statistical evaluation of error metrics, and experimental validation using a 14-bus equivalent of the Yauri substation modeled in PowerWorld Simulator. Results demonstrate that GAN-augmented LSTM reduces Mean Squared Error (MSE) by 38% and improves prediction accuracy compared with baseline machine learning models. This study underscores the potential of artificial intelligence for enhancing grid reliability and provides a foundation for intelligent decision support tools in the Nigerian transmission network