In line with EU, UK Government and JISC policy, Spoken Word Services has a commitment to using Open Source software wherever possible. Our production and development servers run the Apache HTTP server and the Apache Tomcat Java application server on a variety of GNU/Linux and OpenSolaris platforms.

We use a ‘Separation of Concerns’ (SoC) model to ensure consistent, robust access to our digital audio resources. We have assembled 3 functionally separate layers:-

1. Banks of content: the ‘essence’ (primary audio repositories)
2. Catalogues and ‘finding aids’ (secondary and tertiary repositories)
3. User Applications (the ‘presentation layer’)

For this model to function correctly, rich, high quality metadata is essential. Resources from the BBC archives contain metadata describing the content and circumstances of their recorded programs in a native BBC schema. Spoken Word Services have developed field mappings from the BBC structure to Dublin Core metadata. Each item then contains a full native metadata record for description as well as a Dublin Core record for indexing and searching. Additionally each digital encoding of the audio is accompanied by technical metadata describing the encoding parameters and digital file formats. Structural metadata is needed to package these various audio formats, descriptive, technical, and administrative metadata, and written transcripts (where available) for each recording.

This tight bundling of various kinds of metadata and written transcripts with various digital encodings and formats of the audio recordings to encapsulate a single item presents a challenge for content management. Managing these complex objects and maintaining versions of the various files that comprise each item is accomplished with a digital repository. In our production environment, Spoken Word Services are currently using a repository product called REPOS, developed by our partners at MATRIX, Michigan State University. This is publicly accessible via the Padova repository front-end.

Concurrently, and in collaboration with our partners at Northwestern University we are developing our use of the Fedora Commons repository architecture. Fedora Commons is an open-source digital object repository funded by the Mellon Foundation and primarily developed by Cornell University and the University of Virginia. It provides a flexible and standardised object-oriented framework for building and maintaining and accessing collections of digital objects.

Our user applications for audio annotation are described in the Using Audio » Annotation section of the site.