Streaming music if artificial intelligence sings: worries for artists – Corriere.it

It seemed like a joke, but a few months later the phenomenon exploded: music songs created by artificial intelligence (AI) with the cloned voices of famous singers or deceased artists, such as David Bowie, Freddie Mercury, Michael, are being circulated on social networks. and streaming platforms Jackson. The first hit that made a splash was “Heart On My Sleeve”: it seemed that this was a new name for the Canadian rapper Drake, but it turned out to be a fake song. The AI-generated song, narrated by Drake, was launched in April on TikTok by Ghostwriter977 and went viral, racking up 230,000 plays on YouTube and 625,000 on Spotify.. Until the artist threatened legal action and his record company intervened in May and made the song disappear from Spotify, YouTube, Apple Music. However, a month later, the same song, plus a dozen more algorithmically generated songs featuring Drake’s voice, topped the Ai Hits chart. It’s a site created by Michael Sayman, a 26-year-old computer scientist, who reviews the 100 most listened to songs created by artificial intelligence online, including songs with the cloned voices of stars like Travis Scott or Rihanna. “We must fight to protect our human capital in the face of artificial intelligence,” Sting told the BBC.

Between judgment and curiosity

But opinions in the music world are contradictory. Some artists are experimenting with artificial intelligence tools to make it easier to create new lyrics or compose lyrics. “The revolution brought about by generative artificial intelligence in music is comparable to the Napster revolution and is the most important since the advent of file sharing and downloading,” says Enzo Mazza, CEO of Fimi (Federation of the Italian Music Industry). Today the music industry does not intend to resist this, it is trying to manage and integrate this technological breakthrough. And also because generative AI is now part of music production. Those who use an app like Boomy can create an instrumental backing track for vocal recording in a matter of seconds. Thus, users created 14.5 million songs.” BoomyAi has already generated 14.5 million songs. And since January, Google Music ML has been composing music based on lyrics and imitating the voices of performers.; OpenAi Jukebo, the music equivalent of ChatGpt, introduced on April 30, generates music of all styles on demand; SongStarter from BandLab allows you to create an instrumental composition based on lyrics and emojis. And Reactional Music, designed for video game music, generates songs based on an artist’s data set.

Youth trends

Generation Z, which has sharply increased streaming music consumption (which in Italy grew by 16% in the first six months of this year), loves not only to share, but also to invent new music mixes. A survey conducted this summer by Fimi in collaboration with the innovation center Giffoni and Città della Musica in Naples reveals new views among nearly 3 thousand users, 70% of whom are under 34 years of age. To the question “Would you use AI to create music?” 10% of people under 34 years old say they have already done this. 37% believe that in the future artists will be replaced by artificial intelligence, although at the expense of creativity. And another 20% regularly attend concerts in the metaverse.. “For young people, music is a dynamic experience, and virtual concerts offer more participation and interaction with artists,” says Matteo Camarada, 21, a metaverse reporter for SecondStar. And AI is an opportunity to democratize access to musical creativity.” Business threat? Universal Music asked Spotify to remove thousands of songs created by Boomy and submitted by users because they used samples of copyrighted songs.

Possible scenarios

However, this phenomenon is difficult to stop in the face of the wave of songs created using generative artificial intelligence tools introduced this year. “The problem is bottom-up, with AI developers illegally using the vast repertoire of copyrighted music to train algorithms,” says Luca Vespignani, CEO of Dcp (Digital Content Protection), which monitors copyright violations online. Solutions? “First of all, transparency: consumers must be able to distinguish between AI-generated content, which must be reported as such,” says Mazza. And works not created by the human mind should not be protected by copyright or used for commercial purposes. Then you have to use metadata, codes indicating the origin and rights of each piece of music.” Since August, writes Financial TimesUniversal Music and Warner entered into negotiations with Google and OpenAi. Vespignani says: “The goal is to license the use of covered repertoire to train artificial intelligence systems. But cooperation between streaming platforms is necessary to remove unauthorized songs and prevent artificial intelligence systems from exploiting online music.” There is a first agreement: it is dated September 6 between Universal Music and Deezer. The streaming service will reward professional artists: those with more than a thousand streams per month will be paid double.