Here’s how Mureka works.
After I input the “lyrics,” I make a series of choices.
-
First is reference—a big list of song styles. I pick the one that matches the poem.
-
Next, I select from about 20 different singers and styles.
-
Then I choose genre, mood, and vocal gender—though these seem to overlap with the previous step, and I’m not sure whether the new settings replace the earlier ones.
Once that’s done, I hit create.
For Belize, I got two versions. One was clearly better, so the choice was easy. That’s when I decided to go with the standard subscription.
Next came M&M, a long poem that wasn't perfect for a song but, again, two versions—one stood out, so I went with it.
Here’s where things got interesting. With Where Has Suzy Gone? (inspired by Poe), the first version stopped abruptly but matched the mood perfectly. The second had a completely different vibe, so I kept it as an alternate.
Then came I. This time I ended up with five versions, which led me to upgrade to the Pro edition. The first attempt had just the right feel, but it cut off too quickly. Pro allows an extend option, but it asked for more lyrics. ChatGPT suggested I add “[instrumental],” so I repeated the last three lines and included that. Instead of a short ending, I got over two and a half minutes of extra music—so I kept it as the instrumental version and used another program to add a fadeout to the first version.
Going forward, I plan to edit the poems when needed, and maybe add refrains to make them feel more like songs. We’ll see where it goes from here.
