Charisma detection in human speech is challenging but significant for improving human-computer interactions. Traditional methods of charisma assessment are often subjective and inconsistent. Many existing studies on charisma detection using computational approaches focus on numerous features, leading to performance and model evaluation complexity. This study introduces a novel solution by employing a deep learning-based approach to identify charismatic speech in humans objectively. The proposed model leverages Long Short-Term Memory (LSTM) networks to analyze audio features contributing to charisma detection. Using 32 self-labelled speech samples, the model was built and trained to identify these features. Precision, recall, and F1 score metrics validated the model's effectiveness and efficiency. Achieving an overall accuracy of 87.5%, the results indicate that the model reliably identifies charismatic speech, successfully fulfilling its intended purpose. This study contributes to developing an objective method for charisma detection, offering valuable applications in various domains requiring charisma assessment.