Research & DevelopmentAnalysing media in thecloudAn experiment and a marketplaceTristan FerneExecutive ProducerBBC Researc...
Research & DevelopmentA experiment in using the cloud to processa radio archiveA prototype for the World Service archiveA ...
Research & DevelopmentABC-IPAutomatic Broadcast Content Interlinking ProjectUnlocking media archives by making better use ...
Research & DevelopmentThe BBC World Service archiveA 3-year digitisation project50,000 radio programmes from the past 45 y...
Research & DevelopmentThe missing metadataMissing fieldsIncorrect dataSpelling mistakes
Research & DevelopmentListening machines
Research & DevelopmentNoisy transcriptsto be raised in a crisp and easy gait collar tradition and mystiqueand net bottle w...
Research & DevelopmentExtracting topicsExtract keywords from noisytranscriptsMatch to Linked Data topics fromDBpediaDisamb...
Research & DevelopmentProcessing in the cloud26,280 hours of audio processed36,729 compute hours on “small” cloud machines...
Research & DevelopmentMachines + PeopleArchive Machines PeopleArchive+MetadataExperiencesWeb TV+Radio MobileIMPROVESPROVID...
Research & Developmenthttp://worldservice.prototyping.bbc.co.uk
Research & Developmenthttp://worldservice.prototyping.bbc.co.uk
Research & Developmentcomma – Cloud marketplace for media analysisTSB competition for “Innovating in the Cloud”BBC R&D, So...
Research & DevelopmentMedia analysisTopic generation from textSummarising textSentiment analysisSpeaker identification and...
Research & DevelopmentProblems with media analysisComputationally intensiveHard to integrate with other systemsHard to eva...
Research & DevelopmentMaking media analysis easyAlgorithm providers upload algorithmsMedia owners upload content and choos...
Research & DevelopmentThe comma marketplaceAlgorithm developers; e.g. research departments atuniversities and SMEsMedia ow...
Research & DevelopmentAnalysing media in the cloudTristan Ferne, BBC R&Dtristan.ferne@bbc.co.uk@tristanfhttp://www.bbc.co....
Upcoming SlideShare
Loading in …5
×

Analysing media in the cloud

772 views
771 views

Published on

A talk through two projects that BBC R&D is involved in that use cloud computing for processing media. The first is a case study showing how we used cloud computing to efficiently process a very large archive of media and generate metadata, and the second part is about how this led to us to think about abstracting a service out of it, leading to a general purpose cloud service for analysing media.

Full talk notes at http://www.cookinrelaxin.com/2013/06/analysing-media-in-cloud.html

Published in: Technology, Business
0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
772
On SlideShare
0
From Embeds
0
Number of Embeds
373
Actions
Shares
0
Downloads
5
Comments
0
Likes
1
Embeds 0
No embeds

No notes for slide

Analysing media in the cloud

  1. 1. Research & DevelopmentAnalysing media in thecloudAn experiment and a marketplaceTristan FerneExecutive ProducerBBC Research & Development
  2. 2. Research & DevelopmentA experiment in using the cloud to processa radio archiveA prototype for the World Service archiveA marketplace for analysing media in thecloud
  3. 3. Research & DevelopmentABC-IPAutomatic Broadcast Content Interlinking ProjectUnlocking media archives by making better use ofmetadataTSB competition for “Metadata: increasing the value ofdigital content”BBC R&D and MetabroadcastMay 2011 - May 2013
  4. 4. Research & DevelopmentThe BBC World Service archiveA 3-year digitisation project50,000 radio programmes from the past 45 years3 years of continuous audio500TB of high quality audio
  5. 5. Research & DevelopmentThe missing metadataMissing fieldsIncorrect dataSpelling mistakes
  6. 6. Research & DevelopmentListening machines
  7. 7. Research & DevelopmentNoisy transcriptsto be raised in a crisp and easy gait collar tradition and mystiqueand net bottle westphal mia ballroom with a fifth will one of yourvery well that p. c. set a caustic wet plate is sprint says it twice topurposes again whos addicted across stick is a podium whichstopped at a slow start to the masses of setting up a world andon top was a big nineteen ninety three after a renewed spirit ofthe big dig ,comma off trillo .period when you are unable tocompose and see what its stole to working for a while at theguys when i started the eighth that we teach eighteen hamperand a timeless dave theyd each code for my list tinged yellowand io i had no east p. n. c. and i was a big epic tina afootomara i. q. from kodiak and there was so they become koshershopko misfit and i was a david to compose his teams end andat haas tied to districts in the indian head of i. a. moved to beijing
  8. 8. Research & DevelopmentExtracting topicsExtract keywords from noisytranscriptsMatch to Linked Data topics fromDBpediaDisambiguate using distance withinthe “semantic” space
  9. 9. Research & DevelopmentProcessing in the cloud26,280 hours of audio processed36,729 compute hours on “small” cloud machinesProcessed whole archive in 2 weeks at a cost of ~$3,000Built an API for managing the process
  10. 10. Research & DevelopmentMachines + PeopleArchive Machines PeopleArchive+MetadataExperiencesWeb TV+Radio MobileIMPROVESPROVIDESPEOPLE
  11. 11. Research & Developmenthttp://worldservice.prototyping.bbc.co.uk
  12. 12. Research & Developmenthttp://worldservice.prototyping.bbc.co.uk
  13. 13. Research & Developmentcomma – Cloud marketplace for media analysisTSB competition for “Innovating in the Cloud”BBC R&D, Somethin’Else and KiteMay 2013 - May 2015
  14. 14. Research & DevelopmentMedia analysisTopic generation from textSummarising textSentiment analysisSpeaker identification and diarisationMusic identificationMood classification of audio and videoFace recognitionSegmentation of audio and videoObject and place recognitionScene detection in videoSubtitle creation
  15. 15. Research & DevelopmentProblems with media analysisComputationally intensiveHard to integrate with other systemsHard to evaluate and compareHard to know whats possible and what’s available
  16. 16. Research & DevelopmentMaking media analysis easyAlgorithm providers upload algorithmsMedia owners upload content and choose what they wantto analyseThe platform manages:Computation and scalingStoring the dataMonitoringBilling
  17. 17. Research & DevelopmentThe comma marketplaceAlgorithm developers; e.g. research departments atuniversities and SMEsMedia owners; e.g. broadcasters, museums, archives, evenindividuals
  18. 18. Research & DevelopmentAnalysing media in the cloudTristan Ferne, BBC R&Dtristan.ferne@bbc.co.uk@tristanfhttp://www.bbc.co.uk/rdhttp://worldservice.prototyping.bbc.co.uk

×