The specter of Artificial Intelligence (AI) has hung over the world ever since ChatGPT made it aware of the possibilities. While it can be tempting to think that the integration of artificial intelligence is something that’s going to happen in the future, it’s actually happening right now.
One of the places where this is going on is the music industry. Music – a uniquely human enterprise – is being infused with AI technology. This has been met with mixed reactions from famous artists. But whether or not you believe this is a net positive or negative, you cannot deny that it’s happening.
In this article, we’ll examine just how common AI already is in music, exploring the topics of:
Shawn Everett, a Grammy-winning engineer and producer, said about AI that “as far as songwriting and production goes, we’re on the cusp of a wave of something that I don’t think we’ve really seen, maybe ever.”
Artists have been using tools like rhyming dictionaries and rudimentary lyric generators for decades now, but AI has opened up possibilities that were only science fiction before.
Now, Large Language Models (LLMs) can generate sophisticated lyrics and respond directly to user input. Consider the example below, where I asked ChatGPT to write a love song.
In a few seconds, the chatbot generated the lyrics for a song. While the words are generic, the song is coherent.
Now consider what happens when I give the chatbot a more interesting prompt.
A few more seconds, and ChatGPT takes the first song and morphs it to fit my more specific prompt. In the same way, you could spend more than a single minute and work with the AI like a genius co-writer to hone your ideas and test countless possibilities.
There are lots of other AI tools besides ChatGPT for generating lyrics. The website These Lyrics Do Not Exist can create lyrics in different styles, moods, and genres. Artists are already using tools like these to glean inspiration and test new ideas.
AI in music creation doesn’t stop with lyric generation. There are multiple AI tools that have been created to compose music. These are just a few examples:
An artist, Taryn Southern, used products like AIVA and Amper Music to produce her album “I AM I,” one of the first albums to be created entirely with the assistance of AI.
One of the most famous examples of an AI generated song is “Daddy’s Car,” created by the Sony Computer Science Research Lab in Paris. It was created in the style of the Beatles, with an AI trained on a dataset of Beatles’ songs to generate the melody and harmony, which then human musicians arranged into the final product.
AI isn’t just involved at the creation level – it’s also being integrated into musical performance as well.
One of the ways AI is being introduced into music performance is with artificial duets. AI Duet, a Google program, allows you to play a melody on a virtual keyboard, which then the AI responds with a complementary melody. It’s like a jam session, but with a computer!
Other companies are also working to develop AI music performance. The piano company Yamaha developed a system that can imitate the work of famous pianists, analyzing their distinct styles. Through this program, you can enjoy pieces that the AI interprets in the style of your favorite piano players.
Virtual pop stars have begun to grace the stage as well. One of the more famous examples is Hatsune Miku, produced by Crypton Future Media. It uses AI to sing synthetically once you put in lyrics and melodies. Miku has performed worldwide through holographic projection, even singing with live musician backing.
Of course, not all AI technology is being used to create virtual pop stars. Much of it is being used in a behind-the-scenes capacity, making flesh-and-blood performances better.
Systems have been created to help conduct orchestras, for example. In the University of Southern California, an AI conductor was created by learning from human conductors that can conduct and uniquely interpret pieces.
AI can bring people together. A composer named Eric Whitacre created a program called the Virtual Choir to bring voices from all around the world into large-scale performances synchronized by AI that wouldn’t otherwise be possible.
What if you’re just going to a regular ole’ music performance in person? How can AI be involved there? Well, some venues are using AI to respond in real time to the audience and adjust the lighting, sound, and visuals accordingly. AI’s adaptability allows it to create unique performances every time.
AI is the engine allowing for the detailed features you see in music streaming services like Spotify. If you’ve ever wondered how they always seem to have the right song for you in your “suggested”, then look no further – the answer is AI.
Spotify AI analyzes the listening habits of millions of users and the features of every song, like tempo, key, instrumentation, and mood, in order to recommend the song you’re most likely to enjoy next. AI is also involved in their playlist curation, helping them create the best mix of songs for whatever you’re looking for. Spotify’s AI DJ can provide hyper-realistic commentary between songs it is specially curating for you!
Even the search bar is influenced by AI. Spotify uses Natural Language Processing (NLP), which allows you to search for things like “slow dance songs from the 1980s” and get relevant results, as the AI can crosscheck these queries against the song database.
Artists using Spotify get AI-driven statistics to help them, like their audience demographic and listening habits. Spotify AI also can tell artists the real-time performance of their songs so they know what their listeners are gravitating toward.
AI is revolutionizing how music is taught. Apps like Yousician use AI to give you real-time feedback on your timing, accuracy, and technique. Some other programs that use AI to help you learn different instruments and music theory include:
What if you want to learn more “intangible” skills like pitch recognition, chord identification, and other important listening abilities? AI can do that too, with programs like EarMaster, which trains your musical ear and adjusts the difficulty level to wherever you are at.
Apps like NoteFlight use AI to help you learn to transcribe music as well. It can make sheet music out of MIDI files, or if you’re writing music manually, it will help you use notations and dynamics accurately.
Whatever you’re looking to learn, there are so many apps out there that use AI to streamline the learning process.
The advent of AI in music has presented a lot of interesting dilemmas in the world of copyright, some good and some bad.
Already, AI is being used to detect copyrighted music in platforms like Youtube and Facebook to make sure that the law is properly enforced. AI is also being used for positive gain in the distribution of royalties, carefully tracking how often tracks are being played and ensuring that artists are being properly distributed the royalties they are due.
Though we’re only scratching the surface of this problem at the moment, there are numerous problems with AI creating music. The most problematic one – who actually owns the rights to it?
If an AI program is trained on the voice and work of Artist A and asked to sing in the style of Artist B, then the program produces a piece, whose song is it? Is it Artist A? Artist B? The people who programmed the computer? As technology advances, these dilemmas will need to be ironed out.
While AI is already incredibly pervasive in the music world, the technology is still rudimentary compared to where it will likely be in the (near) future.
At the moment, many of these AI tools are spread out among hundreds of different apps. Producer Shawn Everitt thinks that in the future, these AI tools could be consolidated into one plugin that can create almost anything the user requests.
AI is already usurping the roles of many engineers and producers, and at the current trajectory, it looks like there is even more it could automate. The question is how far will we let it go? Can AI create music all on its own, from the words to the mixing to the singing to the instrumentation? And even if it can – should we let it?
These are the fascinating questions our generation has to struggle with. At the moment, however, AI is an extremely useful tool – already a very common tool as well – and we’ve seen it produce a lot of good in the music industry.
