Project Description
Artificial intelligence has played a huge role in the music scene today. When we listen to Spotify, there’s an algorithm focused on recommending music, from our top mixes to the song that plays next in our Spotify. Inspired by the music recommendations on Spotify, my project will focus on using neural networks to generate a song FOR YOU! The plan is to expose the neural networks to various genres, from Hip-Hop to Classical Music, to best fit your genre.
We will start off this project by doing research in how neural neural networks have previously been used in music generation before. Previous work has been done in this field before, but our goal is to attempt to develop a project from scratch/make previous works more efficient. After doing research, our next step will be to attempt to train a neural network to differentiate and detect different genres of music. If this goes well, the next plan will be to work with existing datasets with instrumentals to test whether a neural network is able to generate a beat or instrumental in different genres based on the given data. Our long-term goal would be to collect larger datasets with songs that include lyrics, to see if our neural network is able to generate music with words.
Along this project, we will also consider the ethics behind our project, questioning who should be able to use this technology? And how has AI impacted other forms of art? As AI becomes more popular, it is important that we take a look into the history and issues of Artificial Intelligence in art, and we will propose solutions to solve such problems.
Project Goals
- Research past projects with Neural Networks & Music
- Collect a dataset with instrumentals and music from different genres
- Teach a a neural network to detect different genres of music
- Train a NN that is able to to generate instrumentals automatically for you
- If possible, extend the previous step so that the NN is able to generate music with lyrics in it.
