Child pages
  • Technical Investigation - Opencast Matterhorn

Versions Compared


  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment: Corrected links that should have been relative instead of absolute.

Technical Investigation - Opencast Matterhorn


See for a formal description of Matterhorn.



Matterhorn runs in Apache Felix (OSGi) and can be made to integrate with Red5. The default workflow for Matterhorn converts uploaded (or captured) files to AVI, M4A, and FLV and stores them on the filesystem. Built-in media browser ("media module") with embedded player. Ability to distribute to iTunesU and YouTube. It appears that transcoding is performed using ffmpeg.

Interesting Features

  • Capture client
  • Scheduled recording and processing
  • Captioning support

Open Questions

  1. How easy would it be to bundle Matterhorn to run from a web application server or as embedded?
    • It is important for adoption that VoV not be too complex so it would be best if all of our components would run under the same server.
    • (12/3/2010 Discussion) OSGi inside war is possible for running under Tomcat, but cautioned that it would be better to run it inside a separate container under its own JVM. It was recommended that the encoding services be run on a separate machine to avoid affecting other applications due to heavy load. Also, it was reiterated that the services that make up Matterhorn can be used independently of each other.
  2. How easy would it be to use different video analysis tools for transcribing, scene detection, etc?
    • We will probably want different tools as our content is not exclusively lectures and we might want different tools (or configurations) for different types of content that would be ingested. It appears that this should be easy to do from a configuration standpoint, but it would be good to verfiy.
  3. What plans are there for Matterhorn to develop these video analysis tools?
    • Matterhorn's 2.0 Vision mentions speech-to-text.
    • (12/3/2010 Discussion) Currently video segmentation is working and speech-to-text is being actively worked on but is not ready yet. These time-based analysis are encoded in MPEG-7 metadata.
  4. Does Matterhorn have a plan for managing preservation collections?
    • It appears that right now Matterhorn has focused on an ingest workflow but not so much on the management of already ingested content.
    • (12/3/2010 Discussion) Matterhorn components described as capture client, engage player tools, and media processing pipeline.
    • More generally, where do the needs of libraries fit in the Matterhorn 2.0 vision?
  5. How much will the Engage UI tools provide features that we need?
    • Matterhorn's 2.0 Vision mentions bookmarking, annotating, playlists, and "study reels" (for exam prep).
    • (12/3/2010 Discussion) Bookmarking to be implemented by 2011 and maybe playlists as well. Engage player has been refactored to have a plugin framework.
  6. The 2.0 Vision statement includes mention of possible integration with YouTube and Kaltura. How is Matterhorn technically planning on achieving this?
    • Previous VoV discussions have included integration of third-party content for playlists so it would be nice to know if they are working on this as well.
    • (12/3/2010 Discussion) Integration at this point seems to be focused on distribution to these sites.
  7. How would we integrate with an external instance of Matterhorn? How would we handle updates and deletions?
    • The issue here is what would be the authoritative source of information or would they be kept in sync?
  8. Would we want to use the workflow service or have our code call each of the individual services separately?
    • Both should be possible but it depends on how exactly we want to integrate or manage the ingest process. The only pitfall here might be that if we don't use the workflow service we either will not be as flexible as Matterhorn or we will need to implement a flexible configuration which might be tricky.

Other Notes

  • Echo 360 is a commercial hosted lecture capture system similar to Opencast Matterhorn but appears to only handle capture and not digitization of media.