One of the most common uses of neural networks is the generation of new content, given certain constraints. A neural network is created, then trained on source content – ideally with as much reference material as possible. Then, the model is asked to generate original content in the same vein. This generally has mixed, but occasionally amusing, results. The team at [Made by AI] had a go at generating Christmas songs using this very technique.
The team decided that the easiest way to train their model would be to use note data from MIDI files. MIDI versions of Christmas songs are readily available and provide a broad base with which to train the model. For a neural network, the team chose to use a Long-short Term Memory (LSTM) architecture. This is a model which is contextually sensitive, which is important when dealing with structured formats like music or language.
The neural network generated five tunes which you can listen to on the Made by AI Soundcloud page. The team notes their time was limited, and we think that with some further work and more adherence to musical concepts such as structure and repetition, it might be possible to generate something a little more catchy.
There are other applications for AI in music, too – like these intelligent musical prostheses.
Right now the copyright holder for “Carol of the Bells” is getting lawyered up.
It’s not so much the arrangement of the song as it is the particular qualities of the midi voice. One I’m assuming is either open source, owned by the researchers, or most likely, liscensed to them by the purchase of the midi synthesizer they chose to use.
Alternatively, the copyright holder of carol of the bells had better have a good legal defense because the melody of “Carol of the Bells” falls squarely into the public domain. Only the lyrics are copyright, and I don’t hear any lyrics in the recordings.
If music with similar stylistic features warranted legal attack, then the concept of musical “genres” would be rendered void. There would be only one Christmas song. I would hope it’s a good one.
– Summer Galu
https://xkcd.com/406/
Away in a Manger.
https://www.youtube.com/watch?v=zj3gFmiE3dc
Not about Jesus indeed, like every other thing we consider Christmasy with the exception of the word itself.
Evergreens, feasts, stars, lights, gifts, the character of Santa, yule/jul, “nisser”, and even the damned date, all stolen to make it easier for people to swallow going to Christ-mass.
Jesus isn’t relevant to Yule, the actual celebration, than he is to Easter.
Now, if only we could get it to compose some hard-core gangsta rap…
Sounds like something an elementary school band came up with.
They need the Trans Siberian Orchestra and Mannheim Steamroller plugins
Dont we already have enough?
Its…. Its terrible….
Just like all other Christmas music then.
If that is the AI supposed to lead to Skynet, I am not too worried…
could end up like the santa claus in futurama. you’ve all been verrrrry naughty!
I’d be worried you’d be locked up by the machines, and play those songs repeatedly in the holding cell you’re in.
Skynet or Skynyrd?
B^)
I’m surprised, its less annoying than most holiday advertising jingles.
If someone would just translate music theory into algebra, a bot might have the tools to make a real song.
blergh ????
David L. Parnas–
“I find that most people are interested in artificial intelligence for the same reason they’re interested in artificial limbs–they’re missing one.”
“Artificial intelligence has the same relation to intelligence as artificial flowers have to flowers.”
They had noise on input, so they received noise on output. No surprise there. The real conclusion from this experiment is that the real value of your work does not matter as long as you use strong enough buzzwords.
They all sound the same to me … check
They make me feel a little homicidal after a few listens … check
… yep they’ve succeeded in making an AI that can churn out Christmas noise … erm I mean music