Have you imagined that artificial intelligence is capable of doing something very creative? Of course, we know that AI makes our routine tasks easier. It pushes the boundaries of machine-enabled functionalities in different industries. For example, AI can create human faces that look 100% real. But is it capable of composing music, for instance?

It came true because of AI music generators. Now AI can write music, and if you hear it, you can hardly tell if the computer created it or a human being. So let’s dig deeper to figure out the future of sound.

How do AI music generators work?

Most of those AI music generators rely on deep learning based on analyzing large amounts of data. You collect lots of music tracks, process them, and feed this data to artificial intelligence. Then the training starts. The first results would sound clumsy, just like me torturing a piano at the age of six. But it gets better with a lot of practice and professional supervision.

Knowing that every composition has a recognizable pattern, the process of creating doesn’t seem so magical anymore. Well, at least for the computers. Eventually, learning all the data needed, it can write its look-alike melodies.

And here is the trick. If you use just Elvis Presley’s music as training data, the result will sound somewhat like his songs. And if you use only Aretha Franklin’s albums to explain what music is to AI, then you’ll get something that will sound similar to her.

How does AI sound?

There is an enriched variety of such platforms. Provided with genre, artist, and lyrics as input, they output new music samples. I’ve checked many of them, and here are the three best AI music generators I’ve found.

Jukebox

Apart from others, Jukebox also produces songs with lyrics and vocals in addition to music. What’s more, it goes even further. Together with the OpenAI researchers, it creates music and vocals in the style of Katy Perry or Frank Sinatra. It sounds a little old-fashioned, as if you are listening to your grandfather’s record on his gramophone.

Evoke music

Evoke music allows you to create your piece of music by typing in keywords to a generative tool. All the system needs to know is a couple of things to make your perfect match. First, choose the mood, genre, instruments, and length of the future track. And then you get your tune.

Amper

This platform is similar to the previous one. However, Amper is one of the most reputable and easy-to-use platforms on the market. It builds tracks from pre-recorded samples. Then a user can change the tempo, instruments, or mood of the created song.

FineShare Singify

FineShare Singify is a free AI song cover generator that can help everyone make song covers with ease. You can record your own singing voice, choose one model from over 100 well-trained AI voice models, and convert your recording with the power of AI. Music creators can also benefit from this tool, you can use it to improve your singing voice to a professional level and make amazing music works.

Can AI music generators replace musicians?

We’ve heard about the use of artificial intelligence in music production since the 90s. Back then, David Bowie helped develop the Verbasizer app. It used the words the person typed on a computer and randomly reordered them to create new combinations. So it could help to write lyrics.

Let’s ask David to explain it better:

It was an algorithm based on Bowie’s approach to writing music. And it didn’t work out back then. But now we’re talking about AI that works similar to how the human mind does. So, are we about to replace musicians? Is there enough of what we call ‘magic’?

Robert Laidlow, a British composer who used AI to create music for the BBC Philharmonic Orchestra, is skeptical about the AI-generated music:

“I’ve yet to hear a piece of AI-generated music that is either breathtakingly beautiful or very surprising”, — he says. Laidlow considers AI to be a musical assistant rather than a replacement for human composers.

In fact, these AI instruments open up new opportunities for musicians rather than replacing them. Sometimes it even can be not only fresh but also funny. For example, The Flaming Lips, Deeplocals, and Google’s Magenta team made it possible to play fruits like the piano!

Conclusion

As I’m a huge lover of live sound, such technologies scare me. But I also admit that artificial intelligence is fascinating. AI is made not to seize power or enslave us. Instead, it can be used as a tool for musicians to aid the creative process or compose a piece without much musical knowledge.

Just imagine what a valuable assistant it might be. I’m sure that every musician can find it to be useful. It can be about composing lyrics, adding different instruments, or generating new music lines.

Sound impressive, doesn’t it? But we still need to figure out how to incorporate this into our day-to-day creative process. Meanwhile, people are still making amazing music to convey their feelings and stories. AI can’t do this sort of magic now, so I encourage you to enjoy the music created by humans for humans.

About the author: Lisa Zelenskaya, music producer at Fugue. Passionate about music and media. She explores new genres, styles, and musicians even in her spare time.

Subscribe to
Icons8 Newsletter!