PROD

Export

DOAP Project Details
DOAP
ATOM Project Activity
RSS ATOM
Project Details:

SPINDLE: Increasing OER discoverability by improved keyword metadata via automatic speech to text transcription

SPINDLE will create new tools and HTML5 demonstrators to assist our media team with keyword metadata creation, indexing and display for OER podcasts. SPINDLE will use linguistic analysis tools and Large Vocabulary Continuous Speech Recognition (LVCSR) software to generate a keywords for enriching metadata, will investigate semi-automated generation of full-text transcripts and indexing rich media content using HTML5.
Funding-icon
100% complete
Project Completed
Funding period:March 2012 to September 2012
Source:
Link Organisations:
University of Oxford
Tag Tags:

Standards less

Standard Comments
TTML
WebVTT
HTML5
RDFa
schema.org
LRMI
RSS 2.0

Technologies less

CMU Sphinx
LVCSR
Large-Vocabulary Continuous Speech Recognition
Adobe Premier Pro CS5

Comments:

Comment:
Posted by v1mhawks at 2012-05-25 12:09:08
Looking at using Market Samurai and other SEO tools
Comment:
Posted by v1mhawks at 2012-05-25 12:07:57
planning to examine how google rankings change with inclusion of auto-transcript keywords
Comment:
Posted by v1mhawks at 2012-05-25 12:06:17
reusing tracking techniques used in other oxford podcast pages
Comment:
Posted by v1mhawks at 2012-05-25 12:00:40
Some of the data should be pushed to Jorum
Comment:
Posted by v1mhawks at 2012-05-25 11:54:32
probably be embedding CC license in data files CC-BY-NC-SA
Comment:
Posted by v1mhawks at 2012-05-25 11:53:56
probably be embedding CC license in data files
Comment:
Posted by v1mhawks at 2012-05-25 11:46:41
finding 70% accuracy rate with auto-transcription
Comment:
Posted by v1mhawks at 2012-05-25 11:45:27
aim is to get time coded keywords not a full transcript
Comment:
Posted by v1mhawks at 2012-05-25 11:43:08
re subtitles interested in the workflow from raw to final outputs
Comment:
Posted by v1mhawks at 2012-05-25 11:41:33
issues that captioning standards work very well for video but not as good for audio
Comment:
Posted by v1mhawks at 2012-05-25 11:39:30
adobe premier export as .xmp
Comment:
Posted by v1mhawks at 2012-05-25 11:37:24
'Mining the year of speach' http://www.diggingintodata.org/Home/AwardRecipients2009/MiningaYearofSpeech/tabid/178/Default.aspx
Comment:
Posted by v1mhawks at 2012-05-25 11:36:14
Builds on data mining project at oxford diggingintodata.org
Comment:
Posted by v1mhawks at 2012-05-25 11:31:44
Producing HTML5 video navigation examples
Comment:
Posted by v1mhawks at 2012-05-25 11:30:38
Will be creating a page to showcase steps

Links, feeds, etc

feed Feed
jisc-page Jisc-Page

People:

Person Programme-Manager: Amber Thomas
Person Maintainer: Peter Robinson