Securing the future of music braille production
Hello everyone, I'm Dr. Sarah Morley Wilkins, the Project Manager and User Experience Consultant of the Daisy Music Braille Project. I'm very pleased to have this opportunity to share with you some of our recent strategic development work to secure the future of music braille production and use. Our work will benefit producers and end-users who need to be able to convert, create, explore, and output music in hard copy and digital braille by improving conversion tools and the standards for file formats. My presentation today will be in three parts. Firstly, I'll describe why blindness agencies wanted to focus on music braille and how this project came about, and then share the needs of the sector, what we learned from our international surveys of transcribers, educators, developers, and end users. I'll then talk about how we're addressing some of those sector needs with some strategic interventions, improving file format standards, and then the biggest part of my presentation concentrates on how we're improving two kinds of conversion tools, one for professionals and an interactive user tool. And to close, I'll share what other opportunities exist to further improve the future of music braille production and use. So firstly, why are we focusing on music braille? Well, this project originated in late 2017 when the Norwegian Library of Talking Books and Braille started to investigate what changes they would have to make to support their blind musicians with music braille in future. They quickly realised through discussion that many blindness agencies shared a common concern about ongoing production of music braille, but they all agreed they wanted to continue to produce hard copy embossed braille. Many agencies were and still are facing a decline in expertise. The tools and file formats needed improvement to be fully fit for purpose. And there was a lack of opportunity for effective file-sharing. With funding initially from Norway and more recently with wider sector funding contributions, we set out to fully explore the current concerns and opportunities around music braille through a global sector consultation, and the Daisy Music Braille Project was born. We conducted two global surveys of transcribers, teachers, developers, and end users, and identified a range of factors involved in supporting blind people with music braille resources for education, employment, and leisure, resulting in a State of the Sector report in 2018. Even though digital braille files were becoming more popular amongst users, it was clear that for the study of music, being able to have hard copy embossed braille was still going to be vital. We wanted to make sure that agencies could still produce embossed music braille as well as digital files. We identified four areas of concern which all needed attention to secure the future of music braille production and use. A bit like a jigsaw puzzle, all the pieces need to be in place for the best result. One, input files need to be at high quality as possible for effective conversion. Two, conversion and markup tools need to be accurate and reliable, suitable both for transcribers and end-users. Three, we need good access to existing intermediary files to reduce duplication of effort. And four, we need good teaching, learning, and promotional resources for teachers and learners. The project was immediately able to address the first two of these, the technical ones, the input files and the conversion tools, which I'll talk about shortly. But what about the other two factors we identified? Well, other agencies were already starting to build online collections of music braille files like NLS, The ABC Global Book Service, and Bookshare. So these solutions would make it easier for transcribers and users to find which scores had already been transcribed. And respondents felt strongly that much more could be done to promote the availability of existing teaching and learning resources, and to reduce duplication of effort there. And that's a significant effort was needed to recruit, train, and retain music braille transcribers and teachers. Now addressing the teaching and learning elements of music braille is a big and essential requirement and one which might merit its own project. I'd now like to spend the biggest part of my presentation describing the work we've been doing to address two requirements, firstly to improve file format standards, to improve the quality of the source files to give better conversions into music braille, and secondly, improving the conversion tools themselves so they are accurate and reliable and fit for purpose for future worldwide use. So first, let me tell you more about what we're doing to improve the quality of the input or source files. You all know the phrase, "Garbage in, garbage out." Well, that's just as true for music braille as it is for other kinds of conversions. Good quality music braille relies on starting with high quality source files. There are two main ways of getting music for conversion. The first is to scan an original print score. And the second is to use an existing MusicXML file or to create one. You can scan print music using music scanning and recognition software. Next, you make any corrections and add mark up to fix any internal structure not present in the scanned file. Then you import the marked up file into a music braille conversion tool. The conversion tools usually have a recommended scanning and editing tool such as capella or Sharphigh, which work best with their tool. Now, there are lots of MusicXML files available online, and this should be a good thing, right? However, be careful, some of these MusicXML files are not accurate or complete enough to convert into a usable music braille file. The problem is that some MusicXML files have just enough structure to make it look nice in print or on screen, but not enough structure to tell a braille user what each element is in the school, or it might not be an accurate version of the original. In these cases, the resulting braille file might be entirely useless for learning purposes. We need good files with informative structure. So what are we doing to improve the quality of input files? We're making four interventions. Firstly, we've identified the gaps in the MusicXML standard which currently prevents us from doing a good conversion into music braille. And we've been working with W3C to add the required tags into future versions of the MusicXML standard which would give us the information our sector needs. Secondly, we've assessed mainstream music notation software to establish which tools export the best native MusicXML files for our purposes. And we've made recommendations for improvements. Currently, Sibelius is the tool which generates the best and most complete MusicXML file for conversion into music braille. But as you'll hear later, MuseScore may well be soon awarded that status. Thirdly, we've written guidance for people on how to mark up scanned music files, making sure that the scanned file contains the correct structure and tags to be interpreted correctly by the conversion tools. And fourthly, we've set guidance for music engravers, the people who create print scores and music notation software like Sibelius, Finale, MuseScore. By following our guidelines, they can produce files which are well-structured internally, not just made to look pretty in ink print, and the MusicXML file they create gives us a good conversion into braille. Agencies may wish to outsource to this kind of engraving service in future to create their high quality source files. Okay, now let's move on to what we're doing to improve the conversion tools themselves so they are accurate and reliable. As a result of our international survey of transcribers, teachers, developers, and end users, it became clear that the sector needed two kinds of tools. Firstly, a professional tool for anyone to be able to convert scores quickly and effectively into music braille as easily as possible. And secondly, an interactive user tool for end-users and teachers to write, explore, learn, convert, and output music in accessible ways, including braille. The survey identified the all the existing tools needed improvements to be fit for the future or for wider international use. We reviewed all existing tools and the sector requirements. And through our sector consultation, we created two comprehensive technical requirements documents, one for the professional tool, and one for the interactive user tool, which prioritised all the features and functionality the sector requested together with use cases. These included issues such as accessibility, file handling, formatting layouts, country codes, output options, and so on. We hope to support the development of tools which would be pretty much complete after two years of funded development and which would be sustainable for years after that. We invited developers of music braille conversion tools to respond to these requirements and apply for project funding. A professional production conversion tool will benefit agencies and educators and end-users who need to convert scores quickly and accurately, and if required, through an automated production workflow. The project selected a tool developed by dzb lesen, the German Centre for Accessible Reading, which they built for their own music braille production. And it has many sophisticated conversion rules developed over more than 15 years. MakeBraille, formerly known as Hodder, is an online automated conversion tool accessible through your web browser. MakeBraille is already in use by three agencies in Europe, and ongoing developments are being rolled out after testing. As an online and easy to use conversion tool, it is tightly linked into the score editing and scanning software, capella, so that you can start from a print score, and MakeBraille can now also import MusicXML files as well. MakeBraille can accurately and reliably convert almost all kinds of music scores for education, work, and leisure. And it can be used worldwide as it supports many languages and formatting standards. And it can even be linked into a production workflow if you need to produce music textbooks, for example. MakeBraille strength is that it is an online automated tool, fully accessible with screen access technology. Once you have a good marked up score, the user simply chooses your configuration from a group of design settings suitable for your country, score type, and user preferences, select your file to be converted and then choose create braille. The converted file is emailed within a few minutes, ready to emboss, edit, or read on a braille display. We've also prepared a helpful online support area in English and German with video tutorials, user guides, and guidance for sourcing and creating scores for conversion. So what's coming next? Well, in September, 2020, we began a closed trial to test newly implemented features with 20 testers from agencies in six countries. we've been gathering excellent real world feedback and further country specific requirements. This feedback is helping to improve functionality and our help resources, and shaping this year's development. We anticipate the tool being made more widely available in 2022 at the end of the project funding. During 2021, we'll be adding features including more customization and control of parts, giving users even more flexibility for how they convert their scores. User profiles, easy ways to select configuration options for learners, pupils, teachers, professionals, for example. A setter tool which will format the braille ready for PEF and PDF printing. An HTML5 Viewer and MIDI output, giving users greater flexibility in choosing how they explore their converted score in braille and sound. Contracted braille for countries requiring it with options for handling multiple languages in one file. And tablature, we'd like to try to convert this score type this year. And we'll be adding more guidance for users, for example, I'm marking up specific kinds of scores. So where are we with the interactive user tool for music braille? This kind of tool enables blind musicians themselves to read, write, explore, and output music in braille and other accessible ways. For example, in musical notes, spoken text describing each elements of the score, displayed on screen or printed out in ink print, and importantly for this project, embossed as a hard copy score or displayed on a braille display. Existing tools such as GOODFEEL and BME2 are often used by blind learners, professionals, teachers, and composers who need detailed interaction with a score in several accessible ways, or who are working with sighted musicians or teachers. They offer customization suitable for different stages of learning or study. Our prioritised requirements for an interactive user tool was shared with developers in the summer of 2020, with an invitation to apply for project funding for two years of development. The project steering groups selected related proposals for MuseScore and Sao Mai Braille, which together would result in a fully accessible notation of music braille editing and conversion experience. MuseScore is a leading mainstream, free, open source, sheet music notation programme used by many to create, explore, and export music in sound, print, and even as modified stave notation for partially sighted musicians. But it cannot currently support braille, and blind users have also identified a number of accessibility issues in the tool. Their development will address these issues in MuseScore. By increasing accessibility throughout the tool, improving its MusicXML export, to give better conversion through music braille conversion tools, and to connect with SMB braille functionality which will be embedded into MuseScore itself. Now for the braille side of the interactive music tool. SMB, Sao Mai Braille from the Sao Mai Centre, a nonprofit organisation for visually impaired people in Vietnam, already exists as a braille conversion tool. And a MusicXML reader app also exists for blind users to explore MusicXML files in accessible ways, including braille. Their project will add music braille conversion to their existing conversion tool, applying all music braille rules. It will be able to convert both MusicXML files and MuseScore files in particular. It will run both as a Windows program and as a web service, available directly or within MuseScore, allowing both Windows and Mac OS users to benefit from the tool. They will also be adding a live braille preview into MuseScore. So users can explore the score in braille alongside the other outputs which MuseScore already provides. And will also add 6-key braille input to MuseScore so you can type braille in directly. These two development projects run from January, 2021, to the end of 2022. The combined developments of MuseScore and SMB will make it possible for blind users to have a fully accessible interactive mainstream music notation and editing experience, whether using Windows or Mac OS. So to finish my presentation, I thought I'd share what other opportunities exist to further improve the future of music braille production and use. We're aiming for a situation where more people can get their hands on music braille more easily and in a timely manner. I've already talked about the need to increase the number of trained music braille transcribers and teachers as this is reported to be a problem in most countries. These skilled experts will of course always be needed. Not all scores can be automatically converted, some scores are still quicker to convert manually, and experts will always be needed to transcribe, teach, train, proofread and agree standards for music braille. We also need some concerted effort to increase the development and promotion of teaching and learning resources for teachers and end users and their families. So these are more widely available and used, and children are educated knowing that they can study music with braille support. There is a significant need to have online collections of high quality source files which people can trust as being helpful in their music braille creation. Another future development would be to try to rationalise layouts and customization options for music braille conversions, that is, make options as few and as simple as possible, then the tools would be easier and less costly to develop and maintain, and files would be more easily shareable across different countries. We'd like to see trials of music braille on the newest refreshable braille displays as we might need to consider a special kind of formatting in the conversion tools so that the braille presentation is as effective as it can be over multiple lines. Future tool development might be able to include artificial intelligence to improve context recognition during scanning, file import, and file conversion. This could potentially significantly reduce the time transcribers need to spend in correcting or marking up files, and could make conversions faster and more accurate. Finally, it will be important to continue the professional networking and information exchange in this niche sector to develop and share the specialist and expert knowledge. And hopefully with continued efforts across the world like these, we'll soon be in a position to be confident about the future of music braille for young and adult musicians. And that's it, thank you for joining me today. To find out more, visit our project website https://daisy.org/music-braille. And if you'd like to stay connected to the project, drop me a line at musicbraille@daisy.org. Finally, we extend our grateful thanks to our funders. CNIB, Canada. Daisy Strategic Development Fund. MTM, Sweden. Tibi, Norway, Norwegian Association of the Blind, Norway. Nota, Denmark. ONCE, Spain. RNIB, UK. SBS, Switzerland, and Vision Australia, Australia. And thanks to Haipeng Hu, our Music Braille Technical Consultant, our Project Steering Group, and our testers. I hope we'll stay in touch. Thanks for listening.