This paper presents Speech emotion recognition is an important field of study aimed at developing systems capable of automatically identifying and classifying emotions from speech signals. In this research, we propose a deep learning-based approach using LSTM (Long Short-Term Memory) neural networks for speech emotion recognition. The study utilizes a dataset of speech recordings with labeled emotion annotations. The preprocessing stage involves segmenting the speech data into frames and extracting relevant acoustic features like Mel-frequency cepstral coefficients (MFCCs). The LSTM architecture is designed to capture temporal dependencies and patterns in the speech data. Experimental results demonstrate the effectiveness of the proposed method, achieving an accuracy of 85% in classifying emotions such as happiness, sadness, anger, and neutral. The findings indicate the potential of deep learning and LSTM models in accurately recognizing emotions from speech signals. Further improvements and future research directions are discussed, including fine-tuning techniques and real-time deployment. The proposed method contributes to the advancement of speech emotion recognition systems, which have applications in fields such as affective computing, human-computer interaction, and healthcare.