Artificial intelligence (AI) is scary, and there’s a reason for it. It gradually covers all areas of activity and all areas of life.
Music is not spared either, as this experiment by an Algerian engineer shows. Nobody knows what AI has in store for the music industry as we know it today.
Mohamed Mezdour, 30, is a computer and artificial intelligence enthusiast. His experience is having a song interpreted by an artist who has never sung it. And it worked.
The young Algerian chose a famous Libyan song, “Nabqa houna” (We’re Staying Here) by Adel Al Mashaiti, and was able to produce it with an automatic voice that imitated the voice and intonation of an American singer. The robot reproduced the entire lyrics of the song and its melody without music.
The touch of the machine is perceptible only through the sound quality, reminiscent of wireless communication through a walkie-talkie.
The independent Algerian developer tells London media outlet Al Araby that he used an open-source artificial intelligence model called DIFF-SVC. It is a model of deep learning, or deep learning, based on a mathematical concept called the “diffusion process”.
KI: A new era has begun for music
According to his explanations, the model can be downloaded for free, but it must be trained to imitate voices. It can be trained on any sound you want, or on a “ready to use” voice of a famous singer, for example. And that was the work of the Algerian developer, who says he used an American voice without specifying which one.
Mohamed Mezdour says he is happy with his performance, while acknowledging there is still room for improvement. “The results of my experiments were amazing, but there are still some challenges in the current version of the DIFF-SVC model. For example, to get good results, you need to use quality sounds. And sometimes there can be unwanted distortions in the sound,” he tells AL Araby.
“Producing the perfect sound depends on trial and error, but I think future versions will address all of these issues as the project enjoys strong support from the developer community,” he adds.
Above all, this technology will open up new horizons for amateur artists. “It’s a great opportunity for independent and amateur artists to showcase their talent,” he says.
The downside is that no one can say today what will happen to traditional music and its industry tomorrow.
“A new era has begun in the music world that cannot be stopped,” he says simply.