CRA.I.G - 2 Scoops

Here is my first foray into A.I. music making. For this experiment, I used ChatGPT for lyrics. I gave the software prompts like: "Write a song about a couple eating ice cream with drug dealers down by the Mississippi River in the style of Craig Finn." I would add additional prompts and keep regenerating outputs until I got lines I liked. Then, I used ElevenLabs for text to speech. I used an online interview Craig gave as the voice sample for speech synthesis. The software works in such a way that I could not control voice inflection or flow in any way. Hence, the robotic output. I played around with spelling words differently to see if that would affect output to little success. For example, I tried to manipulate how "Mississippi River" sounded by spelling it Missuhh. Sip. Pee. Riverrrrr. Generated outputs were random and often deviated from sounding like Craig. I did text to speech in chunks and often line by line. I then downloaded these audio clips and laid them on top of an unreleased track by my last band The Ronnie Buxtons with some added background vox by me. Then, I asked Neil Weir to help me mix it to make it sound better. For my next experiment, I'd like to have more control over the voice and maybe Auto-Tune is useful for that.
Click Closed Captions for lyrics.

[@KEAHON]

Poetry

[@MrTimskyscraper]

😂😂😂