Sentiment analysis is one of the most discussed topics in natural language processing. A number of researches have already been made on sentiment analysis, and most of the works are on English language text. There are only a few works that have been found on sentiment analysis from Bangla text. Bangla is the seventh most communicated language in the world, so sentiment analysis on Bangla text plays an important role in detecting the opinion and sentiment of Bengali-spoken people about some products, services, or business. There are lots of microblogging sites and social networks where Bengali-spoken people write comments in Bangla texts. In our paper, we have proposed a special version of recurrent neural network (RNN) model, called long short-term memory (LSTM) to detect the sentiment from the text review dataset. In this regard, we have collected a total of 4000 comments from different online repositories. Our proposed model can successfully classify positive and negative sentiments from Bangla text with an accuracy of 84% and precision of 85%.