Slashdot Mirror
Coming soon: Google TV?
An anonymous reader writes "Google, Microsoft and Yahoo are quietly developing new search tools for digital video, reports ZDNet. Google's effort, until now secret, is arguably the most ambitious of the three, the report states. It quotes sources familiar with the plan saying the search giant is courting broadcasters and cable networks with a new technology that would do for television what it has already done for the Internet: sort through and reveal needles of video clips from within the haystack archives of major network TV shows."
This is a chance for Google to do something truly cool in device connectivity. Even the possibility of a Google TV-Gmail-Google Desktop Search-Google Search connectivity would send their share prices soaring.
People are realizing (yahoo, MSN) that large and bloated is not the way to go.
If I'm not mistaken, vast quantities of tv archive, much of it from the "golden age" when people expected their educational programs to be presided over by professors, is in the public domain. I'd love to be able to dig up some early BBC2.
Imagine being able to look up an old Seinfeld, and then watch it for fifty cents. Or the latest Smallville, or ...
If anyone can pull this off, it's Google.
Agile Artisans
TV Guide info and synopsis information is already available. Even a basic search using this information would be very useful to many people. Google engineers may want to add information to their database, but that would require actually wading through millions of hours of bad television. I volunteer to watch all the anime. The good:bad ratio there is significantly better.
AnimeNEXT anime convention
I'm really looking forward to this.
I'm using Google's image-search very often (and love it!) and I could really use video clip-search.
However, considering how well many sites hide the actual video clips (and I'm not talking about porn), I guess Google might face strong resistance from content providers (wasn't there last week a story about a porn website sueing Google over image-search?)
I don't need a signature.
It would be really cool if google made a TV guide search. For example, if I could go to Google UK and type "BBC1 NOW" - google would tell me what was on BBC1 at the moment. It would be even better if it allowed to specifiy a date and time eg "BBC1 24/12/04 22:00". Now that would be really useful:)
On the subject of useful things for google - how about a currency converter? The convenience of being able to go to Google and type "$10 in £" rather than using XE.com would be pretty cool as well.
I'm a little concerned about the future very high speed Internet being used just like television by the masses. The internet has so much more potential for education and free uncensored flow of information, but developments like these might make a lot of people use the the net only as another way to watch TV. That is a smarter way to make them dumber.
Google's trying to bring TV to the Web the same way they're bringing books to the Web
This is a weird way to describe what Google does... if they bring anything to the Web, it's the Web itself!
But here's the best part of the entire article, IMHO:
Google has been working with National Public Radio and others to index transcripts of audio already on the Internet so that clips can be searchable from its news search engine.
Personally, I would use the video search engine only occasionally... but there is an unbelievable amount of high-quality content that NPR provides on its website, going back years -- interviews, shows, projects, special reports, hell, even Car Talk. The radio thing is a real gem, and I can't wait to use it.
perl -e 'foreach(values %SIG){$_="IGNORE";}while(){}'
I can't agree that the digital archives of any major network are in such a state that finding a clip can be described in terms of needle and haystack.
From my work with the BBC, on a project known as Motion Gallery I'd say that video footage already in a digital format is extensively catalogued and mapped to keyword architecture.
I am also aware of at least 4 other digital archive projects within the BBC. Some of these cover the digital storage of newly filmed material, others like Creative Archive are relevant to making historical footage available online.
The needle in a haystack metaphor is really only relevant to archive materials that are not digital and have been stored on tape or film. Then there is an issue around the cataloguing and ease of searching such material. Even so, the BBC has it's own search system known as Infax. Other broadcasters, such as ITN, have already made their text based archive search available on the Internet.
I think Google can certainly bring some interesting technology and approaches to searching video archive content. This could be in the area of better indexing for existing digital archive footage, or perhaps a search aggregation of text based archive systems in much the same way they provide an image search service now.
Can Google overcome the problem of poorly catalogued tape based media archives? In short no. They could however assist organisations to effectively structure their keyword hierachies when migrating to a digital video format.
Brought to you by the author of such childrens' classics as "Some Kittens can Fly!" and "All Dogs go to Hell."
If you read the MPEG-7 proposals, you'll find that there are provisions for searching based on varying granularities of characteristics of the video and audio both in metadata and within the clips themselves.
I wonder if the frameworks that these guys are developing are within the standard, or if they're going on their own to do this to sidestep patent licensing obligations?
In 1999 a system went live at the BBC that allowed them to search metadata on all their video archives.
I mean, ALL. Even the old stuff - 1870s, IIRC.
They had a big (stonking) database which held the metadata (right down to the equivalent of "This image clip is of Princess Diana wearing a blue dress and kissing a baby"), which was extracted and put into a proper document search engine.
A web front-end was created that performed used the document search engine to get a list of results, then did a lookup to the original database to get the reference for the video clip in question.
Said reference telling you where on which shelf of which row of which large shed to go and look for the video tape/film reel/wax cylinder that contains the clip in question.
Darn good system, reasonably good performance, sucky technology (java applet using CORBA to connect to Java server, HTTP to connect to document search engine, JDBC to connect to original data source).
How do I know this? I wrote the darn thing..
Of course, it'll be obsolete and replaced by now.
I hope.
~cederic