Showcases > REPLAY, a next generation podcasting system
REPLAY, a next generation podcasting system
With REPLAY, the Multimedia Services of ETH Zurich are developing a system for the holistic processing of audiovisual content. From the recording of lectures and events, indexing, archiving and searching, to the distribution of content in different formats, REPLAY covers the complete life cycle of an audiovisual production. By offering opportunities to enrich this object while watching (by adding material, providing comments or making annotations), the system turns audiovisual recordings into multimedia learning objects. REPLAY will definitely change the way we produce, handle, archive, and distribute audiovisual recordings.
Description
REPLAY consists of three components: The recording device in the lecture room, the backend, handling all the indexing, archiving, encoding and publishing of the recorded materials and the user interface, providing intelligent access to the audiovisual contents.
PLAYMOBIL The target of developing a recorder appliance is to achieve a single and affordable box that can easily be placed into a growing number of lecture rooms and automatically capture the presenter as well as the presentation's VGA signal. PLAYMOBIL uses an automated camera tracking system to capture the former, capture cards to grab the latter. The current version of PLAYMOBIL is running on a Linux box based on Fedora, but any other distribution is suitable. Besides a system design and a tested setup of capture and grabbing cards, PLAYMOBIL adds software written in python that communicates with the REPLAY backend to receive recording schedules and ingest captured material. Since this communication involves SOAP request and iCalendar feeds, PLAYMOBIL can easily be adapted to other infrastructures. BACKEND REPLAY focuses on few but important core points:
- Automation and Scalability The initial motivation for the development of REPLAY was to allow for large number of lessons to be recorded. Given that human work in Switzerland is not exactly what you would consider "cheap labour", it becomes clear that staff cannot grow linearly with the number of recorded lectures but instead the recording system had to be automated and scale up to a reasonable number of recordings.
- Long-term archival Comparing yesterday's archiving and distribution formats with what is in use today makes it obvious that the area of multimedia archiving, production and distribution formats is subject to constant change. Formats heavily used today might be obsolete tomorrow. This implies that recorded content has to remain as close to raw material as possible in order not to get stuck with tons of recordings in unsupported formats. Distribution formats are created on demand and in response to the then-used requirements in terms of hardware and software platform as well as quality. Using this patterns, long-term archival can be achieved, making the archived contents accessible many years from now.
- Standards As knowledge gets more and more global, the need to share it grows. Papers are being shared for some time now, and corresponding platform-independent, searchable document exchange formats have established (portable document format, postscript). This is not true when it comes to knowledge contained in audiovisual productions. Various, incompatible formats exist as of today, calling at least for standardized metadata formats.
- Isochronic indexation One of the strengths of REPLAY is the ability to index multimedia content along the timeline. This allows a student to search for a certain term and be directed to the exact point in time within the video and makes downloading and scanning hours of video material obsolete.
The REPLAY backend is written in 99% pure Java. Running as webapps in a Tomcat servlet container, the various functionalities provided by REPLAY can either be running on a single machine but can also easily be multiplied and as such scale up to different needs. To give an idea of how the system is composed, the following image shows the components that make up REPLAY: 
Details on what the purpose of the various parts is would exceed the focus of this document. Please head over to the REPLAY documentation to get more detailed information on this matter.
USER INTERFACE The user interface is set to change usage from a passive, TV-style consumption of monolithic entities to the interactive use by providing time-related content-based access, opportunities to judge, comment, and enrich recordings by adding material. Usage will be collaborative with discussions and comments being shared among students – and lecturers. Thus, all parties concerned will be able to enrich audiovisual material towards more valuable learning objects.
Timeline
REPLAY and the PLAYMOBIL are scheduled (and on track) to run in testing mode by September 2008. The first productive semester will start in February 2009.
