Judith Deschamps: Imitating Farinelli's Nightingale 1/4
Apart from the recurrent gender issues in her creative practice, nothing destined the artist Judith Deschamps to be interested in Farinelli. The artist has no medium of choice, but she is not a musician either. Moreover, it is only very recently that she discovered Farinelli - and not, as one might expect, thanks to Gérard Corbiau's famous film, for which IRCAM’s Analysis/Synthesis team had already worked by recreating the castrato's voice - but during her artistic research on the question of gender.
She was immediately fascinated, not only by this voice—hybrid in its sexual ambiguity, holding as much of the masculine as of the feminine, and by its timelessness, with a child's timbre projected by an adult's lung capacity— but also, and above all, by a particular element of his biography.
His career, in fact, was shortened, explains the artist. At the age of 30, while singing for the London Opera, the Queen of Spain invited him to court. King Philip V had been suffering from what we would probably call depression since his teens. But his passion for music led the queen to believe that the castrato's voice could help soothe the melancholic king. Farinelli agreed. The first concert he gave for the king was a surprise: he sang hidden behind a curtain. The king was flabbergasted and asked Farinelli to become his personal singer. Thus, in a nightly ritual, Farinelli sang his favourite arias to the king until the death of Philip V in 1746. Always the same every evening for ten years.
“I have often wondered why he stopped his career to devote himself to the king," says Judith Deschamps. One of the explanations is that his voice had evolved - like all voices. When you think of Farinelli's voice, you immediately think of an angelic, timeless, asexual voice. But even the castrati were getting older! Their larynxes continued to grow until they were thirty, and their tessituras necessarily diminished.”
Among the arias that Farinelli sang every evening to King Philip V, one in particular attracts the artist "Quell' usignolo che innamorato" by the Italian composer Geminiano Giacomelli (1662-1740), an aria in which the castrato imitates the song of the nightingale. It is this song, as sung by Farinelli himself, that she aspires to recreate, or at least to approach, as part of her artistic research residency at Ircam. However, instead of a 'simple' (a relative simplicity, of course) hybridisation between a soprano and a counter-tenor voice, as in Gérard Corbiau's film, Judith Deschamps' project brings together a multitude of other voices, and makes use of artificial intelligence. Or at least machine learning via a network of artificial neurons, fed by a series of recordings of the same tune by voices of various kinds.
"There was never any question for me of proceeding otherwise than by AI. Firstly, because the method used for Corbiau's film worked very well, so there would be no point in doing it again. Secondly, because of all the fantasies carried by the very concept of AI: there is an imagination, which we can sometimes find a little ridiculous, of a technology that would surpass the human body and compensate for its finitude. In reality, AI, as it is today, is limited. And, above all, AI only works well... thanks to humans! Without humans, the machine can do nothing. This is a type of paradox that I also want to highlight.
Judith Deschamps has been working on this project for a long time. In 2018, she sent an email to Nicolas Obin, a member of the Sound Analysis and Synthesis team whom she knew a little from his past collaborations with artists such as Marguerite Humeau and Philippe Parreno, to submit her idea. He put her in touch with Axel Roebel, head of the Sound Analysis and Synthesis team, but, according to him, the tools were not sufficiently developed to embark on the adventure.
Fortunately, less than two years later, the technology had caught up with the artistic dream...
In the following episodes, we will meet the members of the Sound Analysis and Synthesis team and follow step by step the long process of recreating Farinelli's song by AI.
