AI-generated Christmas tunes aren't as bad as you might think

[Tipup]

REGULAR READERS of these hallowed pages will already know that we're a sucker for a bit of AI-generated silliness.

Whilst we usually look to the wonderful Janelle Shane, there are plenty of other companies doing the same. In this case, jewellery company Made By AI, which creates unique pendants using neural networks, has turned its equipment to making Christmassy music.

Here's an extract from the rather nerdy-wordy blog: "We already knew that working with raw audio input would be difficult in all mentioned limitations. Therefore we realized that a path forward was to train on musical notes.

"We limited ourselves to network architecture types of RNNs (Recurrent Neural Networks) or a LSTM (Long-short Term Memory). There is a lot of well written articles and examples of using these types of neural network. To be able to generate good results of longer sequences of text we choose the LSTM model."

LSTMs, as mentioned above, are a special type of neural network that can remember context even if there are a few steps in between, so instead of being a random progression, there's some level of consistency. In other words, a tune.

It's the same technique that allows your Google Home or Amazon Echo to answer follow up questions like "What's the weather today"….. "What about tomorrow?" without having to learn the context as it goes.

The LSTM was loaded with 100 Christmas songs as MIDI files. Chords, rhythms and progressions were all loaded in, and the whole thing configured to create Christmas tunes of its own, with an instrument of choice and a custom length.

There are some generic examples on the website, but the real charm is that you can specify your needs, and a short time later, receive your own, unique AI-generated Christmas tune.

Naturally, we couldn't resist, so here's our official Made By AI Christmas tune:

Yeah, we're not going to be challenging Noddy Holder with it, but it's fun. Oh yeah - and if you fancy a funky AI necklace - go here. μ