Jukedeck Want to Generate Royalty-Free Music for your Videos

Jukedeck are a UK based startup which is composed (ahem) of software developers, musicians and producers, who have developed a way to automatically generate music to suit any project. Their target is video producers, such as Youtube artists, that need royalty-free music to use without the pain of trawling through royalty-free libraries searching for a piece of music that might work. The software aims to let the user start with a given genre or style (e.g. Jazz, Electronic) and tweak the output to customize as needed in order to fit the requirements of the project.

The FAQs describe the process as follows –

How does it work?

You come to our site, choose a style of music, and are instantly given a completely unique soundtrack. If it’s not quite right, you can tweak it or go for something completely new. You can even choose how long the soundtrack should be, so that it’s just the right length for your video – and we’ve got loads more features coming soon that are going to make it easier than ever to get the perfect soundtrack.

They recently won the Le Web Startup Competition which was held in December of last year, and have received first round seed funding so there’s a promising future for this company.

There aren’t too many technical details provided in either the Le Web pitch (which you can watch in the video above – worth it for the performance at the end of the main pitch) or the FAQs. Founder Ed Rex has a background in Composition and says that the underlying engine has a deep understanding of music theory – harmony, melody and rhythm, which helps the algorithm to generate the music note by note as a composer would. In a write-up by Techcrunch they say

This is based on algorithms that understand the building blocks of music and can use these to create their own tunes in the cloud or on a device.

In an interview with Wired he talks about probabilistic models that inform the composition – probably some form of Markov model.

After each note, it makes a decision: based on what’s come before, what should come next? “That’s where probability comes in – it’s a way for the software to choose different avenues,” the 26-year-old says. “You code it in: so, it’s likely that the phrase will be this long and go to this nearby note, and more likely to move to this chord than that. If it were a case of just choosing between different numbers, the music would be random and wouldn’t sound like music.”

No doubt he’s intimately familiar with serialism and the works of Schoenberg, one of the pioneers of algorithmically generated music.

You can sign up today for early access on their homepage, and I’m hoping to get access to try it out soon and will probably update this post when I get a chance to play around with it. There’s also a sample on the website, with more real-world examples posted on their twitter page.

UPDATE 2015-01-29:
I got a chance to play with the beta version and I have to say it’s very slick and the results are great. The options are a little limited still with a choice of track length, electronic or folk, and fast or slow, but then 3 clicks and about 25 seconds later – the introductory email included an apology that it took so long 🙂 – I had a brand new track ready to download. Here’s my favorite so far. Now excuse me while I go add a soundtrack to those old snowboarding videos.

Dark Story (MP3)

Share this post
  , , , , , , , ,


Leave a Reply

Your email address will not be published. Required fields are marked *