AI and Music

 I have been using Suno AI (https://suno.com/) for about a month, month and a half now. I started off with a free membership, but eventually subscribed to the Pro level, $10/month ($8/month if you pay an entire year subscription up front).

For anyone who doesn't know what Suno AI is, it's a song-generating website that uses artificial intelligence to create the music and vocals. All you have to do is put lyrics into the text field and, through the use of meta tags and prompts, tell it what you want to do with those lyrics.

As I have over 1500 poems stashed away, I figured I'd have no shortage of lyricw to use and to date have created around 200 songs. Not all poems work well as songs, but many I wrote as if they were supposed to be put to music with verses, choruses and even a bridge in some instances.

I did modify some of the poems to make them work in song format as well.

After spending some time on Suno, I realized that it really is just another social media site as people follow you and you follow them back, as people like (thumbs up) songs you publish and you reciprocate. It's social media with a unique feature.

If you use the free Suno, then it claims possession of the music it created for your lyrics and restricts it from commercial use. How it enforces that restriction is unknown to me. However, if you pay a subscription fee, it says you own the commercial rights to the music.

A lot of people use Suno as well as other sites like Songer, ElevenLabs, Mureka, to name a few. I haven't looked into any of those.
It would be interesting, though, to see how differently they treat lyrics to songs I've already generated through Suno.

The results I have gotten through Suno vary. Many turned out really good and others were mediocre. Part of that has to do with the instructions the user gives it, but I believe that the AI is taking cues from the lyrics as well.

Suno is not without its downfalls though. Often it just didn't follow my instructions and that is irritating. For example, I had one song that was supposed to be Samson and Delilah going back and forth in rather humorous dialogue. About 2/3 of the way through, the AI reversed the roles so Samson was singing Delilah's part and vice versa. The second example Suno provided was just all male vocal.

Sometimes a song can get rather glitchy with part of it being muffled. This can be due to several reasons according to a Google search I did.

Also, I told Suno that I wanted a drum intro to a song and a walking bass line throughout. It didn't do either.

One way I have discovered to get more desired results is to record myself singing a particular melody or playing a chord progression then uploading these to make a cover. The AI will use those uploads and create a polished facsimile using the chords and riffs in the recording as the basis for the song.

This link gives and example and I was really pleased with the result.

All Many Such on YouTube

There's a lot more I could say about AI-generated music - the legal issues, the competitive issues. But for me, AI has generated new life in what I've written and I'm not looking to commercialize any of it. 

I'm satisfied with what I have gotten from it.