When opera meets AI

One day when you tried to log in to your account, you forgot the password, and were asked to answer several questions so that the system can verify you are the ‘correct user’ – this is a common experience among us in this cyber era, and this becomes a concept that Sven Sören Beyer, Johann Casimir Eule and Christiane Neudecker take on and create a bold and cutting-edge opera Chasing Waterfalls.

Chasing Waterfalls is the highlight of the New Vision Festival 2022. It claims to be the first opera using AI to co-write the work. It just premiered in Dresden in September. The one presented in Hong Kong is the so-called ‘new vision edition’, with the entire set and five singers substituted by the recorded performance in Germany. The original set consists of a huge ladder in the middle of the stage, surrounded by two cubic boxes on each side. The new reduced version diminishes the visual impact of the set. The staging effect is also weakened. For example, when the five singers, representing five different digital egos of the protagonist, sing and stand high on the ladder, looking down to the protagonist (played by Eir Inderhaug) who stands on the ground, implying they are superior to their creator. However, this hierarchy becomes less poignant by the flattened 2D image. However, this edition did strengthen the interaction between the real and the virtual world, blurring the boundary and confusing the audiences about reality.

The core of the opera is the existential questions of human beings in the digital world. There is no clear storyline. It emphasizes on the conversation between the real protagonist with her digital egos, namely Child, Appearance, Success, Doubt and Happiness. These characters remind me of the Pixel animation Inside Out, but instead of being adorable, like those in the animation, these five digital egos are more threatening and frightening, as the protagonist knows nothing about them, even though they are created unintentionally by her. And they know more about their creator than herself since they keep collecting her data and analysing her behaviour.

The opera continues with some familiar scenes which are common nowadays, including the digital egos picking up selfie sticks to take photos of themselves, and hard-selling their social media platforms and asking people to ‘instant follow’. The creative team even projects the digital images of the audience above the stage (audiences were asked to be captured outside the theatre), adding a layer between the stage (virtual world) and the audience (real world), while showcasing their cutting-edge technology.

The highlight of the opera is how AI composes. There is a scene where all the singers lay down on the ladder, giving way to AI to sing. The music and the lyrics are auto-generated by the AI, which is different in each performance, but the content is more or less the same, just using different wording. The virtual sound of the Eir Inderhaug doesn’t merely sound like her, but contains all the subtlety of bel canto singing such as phrasing, articulation and breathing point. This AI monologue is a reflection on its existence which reminds me of the famous “Tears in Rain” monologue from Blade Runner. I was touched even though I knew it was not real. I think it proves the “make-believe” power of music.

Hong Kong composer Augus Lee is responsible for the creation of the live music. He is also the conductor of the nine-players ensemble. Very often the live music calls for thick texture, and a noisy and intense atmosphere, maybe it hints at the tension between the real and the virtual world. It reaches the climax with the singers singing in tutti at the end where they confront each other directly. As a conductor, Lee serves as a bridge to pull the players, the singer and the pre-recorded music together, which requires very precise control of timing. He had done an excellent job with his very sharp and firm conducting gestures to keep the music flowing like a waterfall.