Music is a beautiful thing. Most people are born with the ability to differentiate between music and noise. Music is supposed to be an instant pick-me-up for most people. Music is also known to help improve the brain’s functioning by improving our memory and our ability to understand things, it is also said to be effective for people with autism. Traditionally, music was made by people using instruments (even digital instruments like the synthesizer) or DJs by using a set of sounds. But now we’re starting to see more and more music being made either completely by using AI or by using it with a person to assist the artist.
We can look at how the usage of digital instruments like synths was opposed by groups of musicians to see how technology is usually not accepted in the music industry. Some people were so against it that they tried to ban the usage of synths because synths made it much easier to create the sounds that came from stringed instruments and that was causing some bands to replace people who played stringed instruments with people who played the synthesizer. Now we see ourselves asking similar questions and then some more, more questions because AI is much more complex than a synth.
AI in music streaming
It’s hard to discover new artists. Music streaming apps like Spotify help achieve this by suggesting artists who make music which is similar to the type of music that a person listens to, and this is done by an AI which ‘listens’ to the songs and keeps track of what type of music a particular person listens to. This AI can also make personalized playlists for individuals based on what they listen to. This has really helped smaller up and coming artists to get into the mainstream and this is an encouraging step for more people to get into music.
Problems associated with AI in the music industry
The thing with AI is that it needs to be trained. In the case that we consider, the AI will have to go through a lot of songs just to get an idea of what good music is supposed to be. The problem that arises here is that music is copyrighted. Making an AI ‘listen’ to copyrighted songs would mean that the AI would create tracks similar to the copyrighted songs that it ‘heard’. In a legal sense, it’s hard to understand if the original artist should get any credit for the AI generated songs. Also, should the people who worked on the AI and made it listen to certain songs, copyrighted or not, get any credit for the songs that the AI generates? Who earns the money that’s made by an AI generated song?
According to the current laws, there will be copyright infringement only if the AI generates a song which sounds a lot like an existing song and all other cases are exempted from copyright infringement. Obviously, this seems extremely unfair for the artists whose songs are used by the AI to learn.
I pity the lawmakers who will have to make these decisions because it’s obviously going to make some stakeholders unhappy. Knowing the pace at which AI is growing, the associated laws need to be made pretty soon.
(The article is authored by Nishant Nijhawan, Intern, Techxty.com. Nishant is a student of computer engineering at Delhi Technical University who believes in optimal usage of technology.)