It seems as if my latest journey platform-jumping AI tools was always going to lead me here, to a place where I would “collaborate” with AI on the theme of bird song and music composition, and work to compose and record a piece of music using AI and my own musical skills.
Let me back up, in case you have not been following along with me over the last week or two. This is the fifth post in a series of inquiry dives where I am exploring various AI platforms on the theme of personal curiosity: how music composers have used bird songs to inspire music (sparked, in part, by Atlas Obscura’s newest book).
- In the first post, I used Google’s Learn About platform to, well, learn more on the topic of bird song and music composers
- In the second post, I took some information from Learn About and moved it into NotebookLM, and did more research and thinking
- In the third post, I took the AI voice podcast generated in NotebookLM and cut it up, and remixed it, turning those synthetic voices into more of an audio collage with a repeating refrain
- In my fourth post, I took ideas from that audio collage and asked Suno to write me a folk song
Which brings me here.
For this iteration, there were definitely more steps to take and more involvement on my part, as the human collaborator. I was really curious about how I might put some of the previous inquiry into action, by creating a piece of music that incorporated bird sounds but also with the help of AI collaboration.
The video down below is the result, but here are the steps to my mad process:
- First, I went into Claude AI platform (my top choice these days, for its safety and privacy features, and relative transparency) and asked it to generate a piece of music manuscript inspired by the song of birds. Honestly, I didn’t think it would be able to generate anything useful for me. But, surprise, it did, and in fact, it generated an entire page of computer code that then became a short piece of musical manuscript.
- Second, I realized that I wasn’t specific enough so I asked it to re-compose, with a Wood Thrush as its source. Again, it created a set of coded instructions that generated a few measures of musical notes, annotated with trills and slides and other bird-centered techniques, as far as I could tell. (see image at top of this post)
- Third, I realized I had to find a way to play the notes, but as a bird, singing. So I went and found a downloadable file of Wood Thrush songs, and grabbed it. Then I took that file and went into Soundtrap (my go-to musical recording platform) and used its Sampler tool to create a Wood Thrush sound that I could use as a single note, so that every time I hit a keyboard note, it would sound like the Wood Thrush. Plus, I added some effects.
- Fourth, I played the composition generated by Claude on my keyboard in Soundtrap, with the Wood Thrush as the sound generated by my playing as I read Claude’s manuscript. That was cool, but it still wasn’t enough.
- Finally, I composed my own keyboard and synth background parts to complement the Wood Thrush song, and then layered in some more of the original bird audio, creating an audio tapestry of sorts that, I think, sounds pretty interesting.
And while the final piece is rather simple in its sound, it has a lot of complexities built into the process, and I see it as a creative intersection driven by my own interests, using AI as a collaborator and teacher at points along the way. It seems important to me that I ended up in a place where I was actually making something, instead of just having the algorithms spit results of queries out to me.
Peace (sounds like birds),
Kevin