“Either you learn or others will and they will fire you.” Music Anxiously Monitors Artificial Intelligence Disruption in the Rest of the Arts | technology

“The musicians see that the sky is very cloudy and we are glad that it hasn’t rained yet,” says Berto Rojo, composer and guitarist for Nina Polaka. The dizzying outage of applications created with artificial intelligence in artistic or cultural fields such as illustration, photography, video, writing or education is causing a strange upheaval in the world of music: will the same happen to us? Think artists and producers.

Rojo is interested in innovation and is already using non-generative software in his technical work, for example to improve models. But he thinks the big musical disruption has yet to explode: “It’s early. It’s very different for an AI to understand an image than a song, there’s a lot more information in a song. We’re still a long way from saying to an AI ‘Make me Madonna’ as if Chopin played it.” on the piano.”

In January, Google published a system called MusicLM, which accurately generates music from text. For example: “A mixture of reggaeton and electronic dance music with an alien, otherworldly sound. Its being urges an experience of being lost in space, and the music is designed to evoke a sense of wonder and awe, while being danceable,” is the request of one of the 30-second tunes she created. Google. The application is not open yet. Rogo has heard it and thinks that “if you don’t tell anyone that the AI ​​did it, then the music they can enjoy”, not realizing that it is not the work of a human. This is the tune Google got from this request:

Google’s novelty is not the only one in recent days. famous disc jockey David Guetta did sing To Eminem in concert, although the voice was not the real voice of a rapper but an even voice but it was created artificially. “I am sure that the future of music lies in artificial intelligence. I am sure. Without a doubt. But as a tool.”

What if the music was (sort of) different?

These are just two examples of AI systems chasing music from different fronts: melodies, arrangements, vocals, and lyrics. It’s hard to think of a near future where AI won’t have a role in artistic creation, at least as a “tool,” says Guetta. Perhaps the level of earthquake depends on the intrinsic features of each art. Music has two supposed advantages over other disciplines. The first is that the musician and his audience have a more direct relationship than other disciplines. It’s rare not to want to put a face on a singer or want to see him live. The second is that music already works with a lot of software and has been stuck digitally for years. Some styles have naturally absorbed the creation of melodies from others, such as using samplesor generate sounds using a computer.

None of the experts consulted by EL PAÍS, however, considers these two exceptions to be final or inseparable. The speed of change in the sector is unprecedented and anything is possible: “We can have no idea what the world will be like in five years,” says Berto Rojo. “Either you learn or other people will and they will fire you,” he adds. This speed has creative implications: “We think that the song is more complex, that it requires some other kind of creative process or that it is difficult to achieve an excellent end result,” explains Chiara Hellquist, director of Vevo Spain, a digital music video platform. “But we were saying the same thing when the MP3 came out and people just listened to it worst quality What is vinyl with AI, we will have a larger volume, the speed at which the songs come out will be more intense, but we will see.

If there are now constantly new songs by popular artists, how many songs can be created with the facilities provided by artificial intelligence? “We will adapt. Now I find it difficult to increase the pace of creativity, but who knows, maybe 20 years ago it seemed impossible for an artist to release a song every month. Now it is almost necessary to keep you in the spotlight,” says Hellquist. New, more customized platforms may also appear. Just five years ago, few thought snippets of melody would explode on TikTok. There are now themes with specific 15-second segments that are perfect for the platform.

Another obstacle to the explosion of AI in music is the direct relationship with the creator, something less common, for example, in illustration or in videos. But why aren’t there more “musicians”, artists with little musical knowledge who for some reason know how to achieve success with AI? Composer Berto Rojo sees this appropriation of musical space as possible by people who know how to write requests for the instrument: “This kind of person can generate a community of fans around their music and make a living from it using artificial intelligence. The artist decides what is left and what is The application did not provide it. The other thing is how free they will be able to use AI and how freely they will be able to monetize this music,” he says. Flooding the music market is not entirely ruled out, but will a new platform emerge where we can consume more and more personalized music?

There are authors today who are approaching production more quickly: “New artists appearing in Pop bedroom Or in urban areas, they compose directly using a computer, and the production table is at home,” says Miguel Martorell, Altafonte’s Southern European Director. “The AI ​​applied to these instruments can be very interesting because you don’t need to compose it yourself, But you can directly ask the AI ​​to do it for you and tweak it.”

The role of libraries

There is already a segment of this sector, perhaps the most anonymous or of least creative value, that AI could sweep: royalty-free, pre-recorded libraries of videos or ads. “Now when you upload a video to YouTube, for example, you choose the library music, ‘Nocturne Piano,’” says Martorell. “This standard is still used by a lot of wallpaper creators in their videos. The AI ​​here can be very interesting because each creator can submit some criteria and get an exclusive tune or song.”

These two examples are useful for understanding this kind of natural progression between what has already been done and what AI has to offer. There are already AI apps now that tweak demos to leave them sounding really good: “You have a rock song and you want the different track volumes to be as close to rock as possible,” says Fran Liu, the record company’s director of innovation. “And then this AI, which a lot of rock songs are put into, analyzes how you’ve mastered it, and draws certain conclusions and applies it to your songs.”

Tools like Lalal, Supertone or Bandlab, which are already widely used, approximate what AI will do in its entirety. This example is a synthetic voice-over created using Voiceful from text only, with no other technical requests for tone, pacing, timbre, or post-production. Creating this synthetic voice took about 20 seconds:

These apps make what already exists so much better without the help of a producer and hours of studio time. The next phases will gradually introduce new options that are easier to use, like this explained by Berto Rojo: “I’ve heard of an AI with which you record a bass track at home with a bad sound card, pass it to the AI ​​and you can say, ‘Make me sound like a bass pixie’.” And it is capable. to take your track and convert it to a file professional. This opens many doors because it will be easier to look good,” he says.

The skepticism of some of the more penitential novelties, such as Antonio Luque, of Father Chinaro, also has its musical reasons: “Those of us who make pop-rock songs know that the ingredients we use are few. We don’t use harmonic sequences so complex as those in jazz.” We know it’s all sol do re, ri la sol and ‘Love Song’ and ‘I Saw You Like That’, ‘I Die for You’ and ‘Until the End’ and the sun and the moon. With what we all deal with, some have success and most don’t. It does. Not even a person who has achieved one success is guaranteed to achieve another. If it were that simple, I would challenge the AI ​​to start achieving success,” he explains.

Not everything was original

In other styles, the AI ​​would obviously not be as inventive: “Not all the songs that hit are 100% original compositions. The vast majority of rap is built on samplesHe took a piece of the song, remixed it, looped it over, and put it on top of the other sounds to turn that melody into a whole new song. It’s very common in certain species and has never been a problem,” explains Rogo.

Despite this detail, it is possible, however, that the novelty will be revolutionary for some reasons that are difficult to predict now: “It’s a scary thing. People who live off their creativity may fear that this technology could suddenly do the same thing twice,” says Berto Rojo. Hurry up.” There are two types of artists: those who will hate it and end up using it when everyone else uses it, and those who understand that this is here to stay, that its growth is exponential and that to be competitive you have to learn how to use it. It’s like Excel comes up and you have a company and you say, no, I want to keep using my notebooks,” he adds.

In the end, as in the rest of the arts, all involved insist that they believe humans will be absolutely necessary to give the final creative touch. Or the machines will bore us and we’ll have to choose what’s good. Or even something else: “Artificial intelligence already exists. It’s called culture. It’s centuries of songs. When I want this help, I listen to songs that have already been created and without realizing it, I reassemble what’s already there. If that’s what it does computer, I’d rather do it myself,” says Lockie, adding that it’s time to smile right now: “It’s the whole idea of ​​parody. I see it as a joke, even with the threat of tech companies telling us they can replace us whenever they want, like They did with our supermarket cashiers or those who fill up with gas.

You can follow The Country Technology in Facebook And Twitter Or sign up here to receive The weekly newsletter.