The open source and open standards Annodex technology applies the Web’s architecture to video and audio for truly simple distributed media publishing. Just as it does for text, this architecture allows for arbitrary sophistication in content generation despite a very simple output format. This design allows for both presentation and powerful metadata retrieval.
This session introduces Annodex from a developer’s perspective: to demonstrate, CMMLWiki is a wiki and multimedia CMS that uses AJAX techiniques for editing video metadata on-the-fly, and Annodex for in-browser video playback from arbitrary timed and named offsets. The technology allows the production of video-based web sites that can be efficiently searched and discovered.
This session shows how to use familiar server-side scripting to build dynamic, interactive video webs. Using AJAX, CGI scripting and database backends, you can provide customization and personalization of video content on an unmodified web server. Using normal scripting languages, you can create hyperlinks into and out of video and manipulate tracks of timed XML metadata.
These techniques can be applied to many languages. The session will give examples of server-side scripting for dynamically compositing and querying Annodex media in Python, Perl and PHP. Lower-level libraries for mangling and re-editing Annodex video bitstreams also exist for C, C++ and Haskell.
Achieving the vision of a global, interconnected web of videos is not simply a technical problem; the integration of Creative Commons metadata is crucial to encourage free and open distribution. The software and techniques demonstrated in this session show how users can host and publish personally produced video, and ensure that they do so in a manner that helps build freedom for others.
About Annodex:
Annodex (http://www.annodex.net/) is an open standards based technology that extends the World Wide Web’s hyperlinking, searching, and compositing infrastructure to time-continuous data, enabling video surfing, searching for clips of audio and video files using ordinary Web search engines, and on-the-fly composition of a video on a Web server from previously annodexed clips.
Annodex media is compatible with many existing media frameworks including DirectShow and open source systems including GStreamer, VLC, and xine. It can be browsed with native plugins for Mozilla Firefox and Internet Explorer or using a Java component. Annodex standards have been designed to scale down gracefully to support handheld devices, for which a browser for Symbian mobile phones has been developed.
Apart from the general CGI techniques covered in this session, Annodex media can most simply be served out using mod_annodex, a module for Apache httpd providing timed URI handling and CMML pre-processing.
Conrad Parker is currently based at Kyoto University, Japan, investigating video search infrastructure and interface usability. He previously worked as a Senior Research Engineer at CSIRO Australia on free software and open standards related to Annodex, a web-architected, distributed hypermedia system. He has presented Annodex related work at meetings of the IETF, the WWW2005 Developer’s Day, ACM Multimedia and various Free Software conferences, and published articles about Annodex in IEEE Multimedia and ACM Multimedia Systems.
He has developed some other useful free software projects including the Sweep sound editor used by Pixar in the production of animated feature films. Conrad contributes regularly to Free Software multimedia projects such as Ogg Vorbis, Speex, Theora and xine, helping build technology for all people that surpasses the breadth and impact of television.
He holds degrees in Computer Engineering and Mathematics from the University of New South Wales, and is a former board member of the Australian Unix Users Group and a former president of the Sydney Linux Users Group.