Bizarre and Wonderful first-draft experiments in LLM (generative A)+pitch-class sets voice-synthesizer computer improvised singing
(done in a jupyter notebook):
Prompt sent to OpenAI api ("Tell me a joke")
response parsed into phonemes (syllables) -> converted into IPA phonetic symbols (using eng-to-ipa)
-> converted into Japanese Katakana -> combined w/ random pitch-collection & durations twice
-> written as two musicXML files -> twice rendered by sinsy singing voice synthesizer
-> mixed together as a duet using librosa