It definitely is not a glorified PRNG. The idea is that you can create patterns that have both variety and repetition with them. I don't like the results, generally, but they are not random.
> experiment with Large Language Models (LLMs) a la GPT-4 or Llama3 in the area of music generation
LLMs are/were indeed used for music generation, but none of these results were convincing from my perspective as a practicing musician. Language is just too different from music, so that results of (symbolic) music generation based on DNNs with linguistic embeddings are only good by chance, if at all. Convincing systems like Udio rather use an architecure for music generation as described e.g. in this article: https://towardsdatascience.com/audio-diffusion-generative-mu.... An LLM is only used to interpret text input and map it to musical features, not for the actual music generation.
This conference paper is from 2005, not 2023, see https://link.springer.com/chapter/10.1007/978-3-540-32003-6_...
Essentially it's a glorified pseudo random generator, like many of the algorithmic composition ideas of the time.
L-systems were proposed for music even earlier. Here's a link to an article from 1986: https://quod.lib.umich.edu/cgi/p/pod/dod-idx/score-generatio...
It definitely is not a glorified PRNG. The idea is that you can create patterns that have both variety and repetition with them. I don't like the results, generally, but they are not random.
Correct; here are some papers that cite the paper above in case folks are looking for more: https://scholar.google.com/scholar?cites=6968059276866815065...
I would assume that a lot of people now experiment with Large Language Models (LLMs) a la GPT-4 or Llama3 in the area of music generation.
> experiment with Large Language Models (LLMs) a la GPT-4 or Llama3 in the area of music generation
LLMs are/were indeed used for music generation, but none of these results were convincing from my perspective as a practicing musician. Language is just too different from music, so that results of (symbolic) music generation based on DNNs with linguistic embeddings are only good by chance, if at all. Convincing systems like Udio rather use an architecure for music generation as described e.g. in this article: https://towardsdatascience.com/audio-diffusion-generative-mu.... An LLM is only used to interpret text input and map it to musical features, not for the actual music generation.
L-System music is a pretty good description of gamelan.