Artificial intelligence is already creating paintings, playing chess, designing and composing songs on its own. Will humans soon cease to be creative? We tell you what the music industry may have in store.
Modern copyright free music, like modern photography, has a computerized future. That means musicians, if you can imagine, will no longer need instruments, but will have access to a software music world which composes music itself.
Artificial intelligence is already producing music that can be listened to, but not without human help. In 2012, researcher and composer David Cope created a computer simulator that tried to compose melodies in the spirit of Mozart and Bach. In 2014, François Pachet, head of Sony's Computer Science Lab, introduced the Evans bot, which could create uncomplicated piano melodies.
The most modern project is MuseNet, which comes up with music in different styles. During training, this artificial intelligence was given several sets of data with the question "Which note could go after this one", and if it guessed, a connection was formed, which turned into a melody.
In all cases, the neural network and the algorithms that help it can generate music indefinitely, but it will always only be an attempt to copy material that already exists. At the moment, it makes more sense to call neural networks musical instruments, something like a modern synthesizer, than full-fledged songwriters.
What about the lyrics?
In 2016, the album "Neural Defense" was released, consisting of songs and poems written by the robot. The algorithm created by Yandex employees Alexei Tikhonov and Ivan Yamshchikov wrote lyrics in the style of Yegor Leotov, the founder of the band Grazhdanskaya Oborona.
A neural network was used to generate lyrics, which was taught to write lyrics in the style of Russian poetry. After that, it was shown Yegor Leotov's lyrics, set verse rhythms found in the musician's songs, and the artificial intelligence generated similar works.
In 2018, the Yandex.music team conducted a study of Russian rap lyrics, in which it traced changes in the genre in recent years. Based on the data, the company generated lyrics that could be written by Russian-speaking artists with the help of a neural network. The result was about 13 lyrics, in the spirit of Russian rap.
What changes are in store for the music business in the future?
At the moment a neural network is not an independent instrument. It works thanks to the people who teach it musical literacy and rhymes, downloading ready-made materials into it. For the time being, you don't have to worry about going to concerts not with live performers, but with a computer. Now, with such "inanimate" music, the writing of new tracks is accelerated. In fact, you only have to load the right "code" into the program, which can produce a decent beat. If some artists manage to release a few albums a year, once they have machine intelligence, the number of releases will increase manifold.
Musicians have a fear for their profession that neural networks will displace them, but performers had the same fear in the '70s when the first drum machines appeared. If musicians embraced the novelty then, it's likely that artists now will simply learn how to work with a neural network as a new instrument.
##What will we listen to?
The music of the future is an endless stream, assembled from hundreds of samples on the fly and never repeated. Already now it is possible to replace listening to artists with listening to what a neural network does, based on the listener's personal desires. For example, the Russian app Mubert, in which unique music is created using algorithms and machine learning technology based on embedded samples. The audio stream changes depending on the listener's musical genre, mood, and other parameters.
Generative streams from Mubert or other apps that create music using machine algorithms become the perfect background music: changes, no distractions, no need for any action to switch.
Posted using LasseCash