These days, music is a huge part of everyday life. It is now included in the entertainment. One way of creating new music that has received a lot of research is automatic composition. In this work, we presented a method for automatically creating random music utilizing deep learning approaches, such as recurrent neural networks (RNNs) in Python using the Keras package and also employing the idea of Long Short-Term Memory (LSTM). RNNSdraws a series of inputs and generates another series as output. We receive the item recommendations from neural networks. This framework is intended to run this method, with musical instrument digital interface (MIDI) files serving as the input data and the created musical sequence serving as the output. This model we employ should be able to recall previous knowledge of the musical sequences. We use RNN and LSTM to recall the musical sequences since we need a system that can remember previous sequences and anticipate the next sequence.