#openscore

Exactly one month ago we introduced OpenScore to the MuseScore community, and one week ago we announced it to the world at FOSDEM, Europe’s largest open source software conference. OpenScore is a new initiative to digitise public domain music, including the works of the great classical composers like Mozart, Beethoven and Bach.

Introducing OpenScore at FOSDEM 2017

Getting involved

OpenScore is only possible with your help. We would be extremely grateful if as many people as possible could back the Kickstarter campaign when it goes live in a month or two, or work with us to produce the OpenScore transcriptions. We really want OpenScore to be a success, so to help galvanise the community we will be offering rewards to those who choose to take part. Kickstarter backers will be able to have a say in which pieces get transcribed, and transcribers will be rewarded with PRO membership of MuseScore.com.

If you are interested in taking part then you can sign-up to register your interest here:

Kickstarter backers: https://musescore.org/openscore

Transcribers: https://musescore.org/openscore-transcribe

Answers to questions

The first blog post generated a huge number of comments and questions about OpenScore, and so I will try my best to respond to them here.

Which pieces will be transcribed for OpenScore?

The initial transcription effort will concentrate on a specific set of pieces selected by the Kickstarter backers, with a few additional pieces that we think will be of greatest interest to the general public and the wider music community. At some point we would like to expand OpenScore to all public domain music, but this depends on the success of the initial campaign.

When can we start transcribing?

The transcription effort will begin in earnest once the Kickstarter has been successfully funded and the pieces selected. However, we are currently in the process of creating demonstration scores to show to potential backers, so if you sign up now then there might be something for you to do before the Kickstarter.

How can I submit transcriptions to OpenScore?

Many of you were keen to know about the process for submitting transcriptions. We don’t currently have the resources to check all scores being uploaded to MuseScore.com, so for the time being we will be approaching individual users with pages for them to transcribe.

Can I submit transcriptions I have already done?

You can tell us about any existing transcriptions when you sign up, but we can’t promise to be able to check all of them immediately. We will contact you individually if we think your existing transcription will be of interest to the Kickstarter backers. Other transcriptions may be considered at a later date.

What is the goal for OpenScore transcriptions?

The goal of OpenScore is not to produce the ultimate critical (or “urtext”) editions, nor is it to produce a beautiful engraving. Instead, the goal for OpenScore is to produce digital editions that are semantically accurate transcriptions of the original public domain editions, which will be fine for the vast majority of users. Furthermore, the OpenScore editions will provide a starting point for creators and arrangers to produce ultimate editions of their own. We will refresh the OpenScore editions with each MuseScore release to take advantage of improvements to MuseScore’s layout rules.

How will the Braille scores be produced?

We are partnering with RNIB to get advice about Braille and MSN notation, and we invite Braille readers in the community to offer their feedback on the Braille we produce. (Our advisors at RNIB mentioned that reading the Braille can even be a valuable way of spotting mistakes in the MusicXML that would otherwise go unnoticed.)

OpenScore presentation at FOSDEM 2019

It happened two weeks ago at the Karajan Classical Music Hack Day in Mozart’s hometown of Salzburg, Austria, where the MuseScore team decided to use technology to bring Mozart ‘back to life’. We made use of open-source technology developed by Google called Magenta, based on TensorFlow, which uses machine learning to create art. We used Magenta to train an artificial intelligence to play the piano in the style of Mozart, but to make this happen we needed lots of sheet music in a digital format.

So, how does machine learning work? Let’s get into a little more into detail with an example of creating a picture. If you feed pictures of Monet’s painting into a machine learning algorithm, the algorithm will pick out the similarities and learn Monet’s unique artistic style. Once finished, the algorithm will be able to turn any random picture into something that looks like it was painted by the master himself. For example, a photograph of a dog can be stylized to look as though it were a painting by Claude Monet. Sounds simple, doesn’t it?

Photo from https://magenta.tensorflow.org/blog/2016/11/01/multistyle-pastiche-generator/ 

OpenScore: digital sheet music to revive classical composers

What if we could achieve the same with music?

The first thing that you need to enable this, is music in a semantical data format. That’s one of the reasons why MuseScore is starting a campaign to digitise public domain music, including the works of the great classical composers like Mozart, Beethoven and Bach. This will be done with the help of people in the MuseScore community, who will take part in a huge crowdsourced effort to produce digital editions of this music. These digital scores can be fed to machine learning algorithms to train a musical brain capable of creating new melodies. The more data and the more training time are available, the better the result will be.

W.A.I. Mozart

The impact of this technology was demonstrated on the 13th of April at the Karajan Music Tech Conference in Salzburg. MuseScore’s demonstration, titled “Jam with Mozart”, is based on A.I. Duet and allows the user to play a piano duet with an artificial intelligence Mozart. The user plays a short melody on a piano, then the computer analyzes it and responds with a short musical phrase in the style of Mozart. The computer was connected to a Bösendorfer self-playing piano, so you could actually see the keys of a real piano moving as though the ghost of Mozart had returned to play them.

Watch the quick demonstration below or try it yourself online!

The New Museum and Rhizome jointly inaugurate Open Score an annual symposium that will explore the state of art and technology today. Convening luminary artists, curators, researchers, and writers to discuss how technology is transforming culture, the first edition of Open Score will consider how artists are responding to new conditions of surveillance and hypervisibility; how social media’s mass creativity interfaces with branding and identity for individual artists; how the quality and texture of art criticism is evolving in a digital age; and what the future of internet art might be in light of a broader assimilation of digital technologies. Supported by the Robert Rauschenberg Foundation, Open Score: Art and Technology 2016 will mark the fiftieth anniversary of the groundbreaking initiative Experiments in Art and Technology. The conference’s title is taken from Rauschenberg’s live performance Open Score during one of E.A.T.’s most iconic events, “9 Evenings: Theatre and Engineering.” Art in an Overseen World Speakers: Simone Browne, Associate Professor, Department of African and African Diaspora Studies at the University of Texas at Austin; Adrian Chen, writer and researcher; Rob Horning, writer, Marginal Utility, and Editor, the New Inquiry; and Emily Segal, Cofounder, K-HOLE