This research develops a comprehensive framework for predicting stock prices using historical financial records complemented with the sentiment of social media on platforms such as X.com (Twitter) and Reddit. We use the Long Short-Term Memory (LSTM) networks to predict time series and the VADER algorithm to measure sentiment. This model incorporates the shortcomings of traditional models that solely rely on historical price records. Additionally, we optimize the LSTM model parameters using Particle Swarm Optimization (PSO) to enhance accuracy. We also demonstrate the superiority of the multi-modal method over other standard methods such as linear regression, XGBoost, and random forest. These findings emphasize the importance of developing models that are able to fuse sentiment expressed in texts and stocks with the sentiment cubed with financial metrics. This is essential for the development of effective forecasting models intended for the financial markets. A potential future development could involve real-time sentiment analysis and the supplementation of data with news articles to enhance these forecasts.