Indonesia’s television industry faces fierce competition, particularly in chasing ratings and ad revenue. This has ultimately led to declining broadcast quality on some national TV stations. This research aims to understand perceptions towards content quality by focusing on public opinion through sentiment analysis of social media (Twitter) using Bi-LSTM and Word2Vec methods. The research involved data collection, preprocessing, vectorization, data splitting, model training and testing, evaluation to find the best model, sentiment data classification, and finally, sentiment data analysis. Using a dataset of 515,492 sentiment points, the model achieved an accuracy of 96.4%, precision of 72.1%, recall of 72.0%, and f1-score of 72.8%. Analysis of Twitter user sentiment leans towards neutral and positive perceptions. The results of the sentiment analysis of Twitter users tend to be neutral and positive. The results of the public satisfaction trend show a change in the pattern of public satisfaction with the quality of television station content.