CSE 190: Music Generation Final Project
We tried to generate Lo-fi music by using neural networks. The architecture that we use are LSTMs. We tried to implement the looping nature of Lo-fi music and generated music with and without accounting for loops. The major finding of our paper was that although the looping nature makes the music sound better, there is a higher chance of running into a cycle. So, there is a trade-off between implementing the looping nature of songs. Finally, we put a drum loop over the best-sounding music that our neural network could produce.