2023-06-11 22:39:39
The fact that a technology can “resurrect” a deceased artist is usually a reason for many media outlets to rush to cover the news with a celebratory tone. Anything that seems to get us closer to a sci-fi movie writer’s wet dream turns the journalist into a zombie following the flute tune and ending in a headline like “AI Creates Songs by Amy Winehouse, Nirvana, or Jimmy Hendrix.” And the truth is that in some cases it is difficult not to fall into fascination.
“This is fucking creepy but amazing at the same time,” reads one of the comments on YouTube. This song is part of the Lost tapes of the 27 club project, an album created using artificial intelligence by the Canadian NGO Over the Bridge, which is dedicated to assisting musicians with mental health problems. The purpose of the album was to raise awareness about the death of young musicians who had problems not knowing how to deal emotionally or psychologically with their careers.
Beyond this laudable initiative that captured the attention of many media in 2021, precisely because of the possibilities that artificial intelligence technology seemed to highlight, many people in the world of academic research are calmly observing these phenomena with analytical tools and helping us. to relativize them.
“Perhaps you can recreate the voice of an artist who no longer exists. But you could also imitate it without artificial intelligence,” says Frederic Font, coordinator of Freesound and researcher in technology and music at Pompeu Fabra University, who is presenting this week the panel on music and AI that has programmed the Sónar festival in Barcelona. In fact, in Lost tapes of the 27 club the part carried out with technology are the instruments and melodies. The voice belongs to singers specialized in imitating deceased musicians and the work process is full of human interventions. Why are we so obsessed with this topic now?
“There is a bubble. We are guided a lot by headlines or stories that seem to explain things that could potentially change the world. I don’t know if it is because we live in a time of existential crisis. And although we are in a moment where many things are happening, neither There have really been so many changes and in the end the real impact on society and on musicians we don’t really know either because hardly anyone is interested”, Font comments.
When Font elaborates on the idea that there are not so many changes for musicians, he compares it to the emergence of sampling: “This has already happened in the past. The introduction of samplers allowed you to use other music already made and add other things on top. I think that we are in a similar process and what seems shocking is that a machine does it.” In fact, we’ve been living with the idea of ’sampling’ in music for 40 years now and the idea shared by DJ and music producer Mark Ronson doesn’t even seem to have expired when he said: “Artists like the Beastie Boys or De la Soul created albums using decades They sampled pre-recorded music to create sonic masterpieces and they didn’t sample because they were too lazy to write their own music, they didn’t sample because they were too lazy to write their own music, they didn’t sample those records because there was something about them that spoke to them and they wanted to. insert themselves into the narrative of that music. And they were able to do that because they had access to the technology that made it possible.”
Font is critical of the terminology and part of the enthusiasm generated: “I think that what we call artificial intelligence, the techniques we use today are not intelligent at all. In other words, the name is wrong, we already started there, no? So what they do is learn probabilities?” In fact, just like the bathroom whose light is automated and ‘smart’ turns off if you stop moving because it doesn’t conceive that you can’t relieve yourself in an inert position, what we call ‘artificial intelligence’ in music can be subject to cultural changes that break the logic of the machine. Using freesound.org as an example, a project he knows well and which is a sound archive: “The fact that in your database the sound of rain is associated with sadness shows that there is no intelligent reasoning there. Now for example that We had been dry for many months, it has rained a little in recent days and the rain has really been something very happy. We can expand the capacity of the algorithms, but they cannot reason.”
Another interesting aspect is the reaction of the industry. Lately, Spotify seems to be moving in various directions that may seem contradictory, but may not be. On the one hand, it announced that they had deleted thousands of songs that had been generated with Artificial Intelligence in a clear gesture to protect real artists. At the same time, last year it acquired Sonantic, a London startup whose technology makes it possible to simulate a human voice from text in a very realistic way. For Font, Spotify just wants to maintain its position of power in the market and its usage data gives a rather graphic idea of what is happening with the majority of artists: “On platforms like Spotify there is a lot of content and if you look at the distribution of “The ones that play the most and the ones that play the least, you’ll see that there’s a very small percentage of the content that plays a lot and all the rest that plays almost never. In the end it’s like the distribution of wealth.”
In fact, for Font it is clear that companies do not hide their interests and in that technology and the debates on artificial intelligence perhaps generate a noise that does not allow us to focus the discussion on that inequality that Font points out: “Specific consumption is very guided by the playlists that the companies themselves generate. Sometimes with automatic algorithms, sometimes not. There is a commercial interest there and many factors beyond the technology itself.” If an established and privileged artist like Snoop Dogg complains about streaming platforms, what wouldn’t the millions of small and independent artists for whom it is so difficult to gain visibility, generate remuneration in the current system have to say?
Font thinks that the big platforms are not interested in certain possible uses and that they would generate a greater distribution of visibility or remuneration: “What technology should allow is that you can search that part of the catalog that you would never reach otherwise. There are a number of possibilities that could be explored but in the end I don’t know how interesting they are for these platforms because they have so many users and they have to maintain such a large structure that they can only optimize for standard users and can’t develop things for that perhaps they would have little use because it is not profitable for them”.
So is there any kind of optimism with these technologies? Font lists several things that seem interesting to him, from “the automatic description of content”, also “that the user can go much deeper about an artist and find out details that now remain almost hidden”, the democratization of access to music “these These tools are going to make creating music easy and really fun, beyond the professional sector I think they are going to offer spaces for entertainment”, in education “because you can play something and an application tells you if you have done it well or badly, or it can even tell you where you went wrong. These tools are going to get a lot better with artificial intelligence.” Font ends with a humanist message: “In education, I don’t believe that any technology will generate a total paradigm shift, make teachers disappear and replace human interaction.” And he thinks the same of the music industry: “I think we have to relativize what is happening, put it in the context of a technological evolution.” Because for Font, there is no technology that can replace the human part: “In the end, I believe that all the personality of the artist, the connection with the public, the link with the community, the discourse…all these types of things are still necessary and they’re not going to be replaced by those kinds of tools.”
#fascination #music #intelligence #learn #probabilities