This week I have been playing with Udio, a generative artificial intelligence model that produces music based on prompts. Its free beta version was released publicly on April 10, 2024. Overoptimistic I started with the prompt "a traditional second-mode plainchant gradual on the Latin translation of two crucial lines in the Dutch national anthem". That did not work.
I tried several other things and ended up with many simple prompts on the first lines of Psalm 129 in Latin. I learned that it is not (yet) possible to produce any meaningful Gregorian or other chant. Flamenco and house sound more convincing, as do classical music and pop. These genres probably work better because they simply provided more training material.
The image in the video is one of the drawings that Peter van Hugten (1949-2023) made for my book “Calculemus et Cantemus, Towards a Reconstruction of Mozarabic Chant” (2015), in which I presented 50 computationally generated melodies for the lost chant of the Mozarabic rite, some quite convincing, others not. Details at: http://www.gregoriana.nl/
Replies