MELD – Music Encoding and Linked Data – is a flexible software platform used by several academic research projects to combine digital representations of music, such as audio and notation, with contextual and interpretive knowledge in the Semantic Web.


It can be helpful to think of MELD as a framework, as illustrated in the figure below, consisting of:

  • User-facing web applications – MELD apps – which run in a web browser and incorporate specific visualisations and interactions for particular music-related tasks and research. Since areas of application covered by MELD apps are varied, they give the broadest indication of MELD’s overall capabilities. Read more about Apps and Projects.
  • MELD Clients Core, a React/JavaScript library providing functionality which is common to many MELD apps, including MELD web page elements, graph traversal actions, rendering and interaction handlers, and state reducers. Our development model gradually migrates generalised functionality from MELD Apps to MELD Clients core as it stabilises. MELD Clients Core reuses and extends several existing libraries; notably, engraving of notation uses components of Verovio.
  • MELD Web Services, a set server functionalities for MELD apps requiring dynamic annotation and session services. From MELD 2.0 onwards, Solid Pods are our recommended implementation for annotation storage; and any provider should be compatible. Since MELD uses web standards wherever possible, implementations can conceivably be seamlessly replaced (or extended) by alternative service implementations. Standard services for hosting web-based music and linked data resources (i.e. non-MELD-specific) aren’t considered part of MELD Web Services.

Behind the scenes in a typical MELD app, users and software agents create semantic annotations of music concepts and relationships, which are associated with musical structure specified by the Music Encoding Initiative schema (MEI). Annotations conforming to the Web Annotation data model are expressed as JSON-LD, allowing alternative music and use-case vocabularies to be applied, to support e.g., popular vs. classical music structures, or rehearsal, performance, or analytical applications. The same underlying framework retrieves, distributes, and processes information that addresses semantically distinguishable music elements. Further knowledge is incorporated from external sources through the use of Linked Data, which is also used to match annotation types and contexts to rendering actions displaying the annotations upon the digital score.


Our current preferred citation for the overall MELD framework is: https://ora.ox.ac.uk/objects/uuid:945287f6-5dd3-4424-940c-b919b8ad2768


MELD overview


MELD exists due to the dedicated work and support of the research teams, projects, and funders detailed on the Apps and Projects page, including significant support from the AHRC, EPSRC, and EU Horizon 2020.