Slashdot is powered by your submissions, so send in your scoop

 


Forgot your password?
Close
typodupeerror
×
Television Media

Searchable Audio/Video Technology 149

Posted by michael from the locate-oprah dept.
wyldchild37 writes: "Business 2.0 has an article on an interesting new technology - TV That Works Like the Web. A new startup wants to make all television content archived, indexed, and searchable."
This discussion has been archived. No new comments can be posted.

Searchable Audio/Video Technology More

Searchable Audio/Video Technology

Comments Filter:

  • Great! (Score:3, Funny)

    by A_Non_Moose ( 413034 ) on Saturday January 05, 2002 @02:34AM (#2789841) Homepage Journal
    now I can get an error 404 on my TV.

    Just what I wanted.

  • Not Again... (Score:3, Interesting)

    by krmt ( 91422 ) <therefrmhere AT yahoo DOT com> on Saturday January 05, 2002 @02:39AM (#2789849) Homepage
    Didn't we hear about interactive TV before? Isn't that garbage over? Granted, TiVo is fairly popular, and it deserves it, but everyone I know wants to sit in front of their TV and be a vegetable. That's what it's good at, and that's what people use it for. This one'll sink because people would rather be lazy.
    • This one'll sink because people would rather be lazy.
      Interactive TV is in its infancy at the moment, and (here in Europe at least) the various stations are still vying for the Most Inane Use of the Red Button on a Sky Remote award (my nomination: vote for the brainiest contender on Discovery Mastermind). And, as you said, people are lazy. However, I suspect this will be a much-used aspect of future TV watching (or 'participating', as our chums in the media will doubtless prefer).

      Imagine for a moment that you're a normal muggle. Someone tells you that one of the cast of Friends has a new celeb boyfriend. You turn on the TV, type in the name of that woman who was on The Misfits of Science, and voila! You need never read a proper news story again.
      • Imagine for a moment that you're a normal muggle. Someone tells you that one of the cast of Friends has a new celeb boyfriend. You turn on the TV, type in the name of that woman who was on The Misfits of Science, and voila! You need never read a proper news story again.

        Imagine for a moment I am a normal muggle with both a TV and a computer. - Can't I already do this online?

      • Re:Not Again... (Score:3, Insightful)

        by krmt ( 91422 )
        Why not the computer then? The TV limits what people can do necessarily. I can't really imagine anyone wanting a keyboard for their TV any more than I can imagine someone buying one for their gamecube. Note that these peripherals don't sell well on systems like video game consoles that are inherently interactive. There's no way someone is going to use one on a machine that is naturally the exact opposite of interactive.

        Producers have spent decades making TV the exact opposite of interactive, and they've done an admirable job of it. I know of no better way to kill your mind than TV, even drugs don't do the job as well or as easily. Then they tried to take this philosophy to the internet with Push technology. Remember when that was the thing? How often do you hear about those guys now? The same way push failed for the internet, interactive TV will fail because it's the antithesis of what people know and love about television. It's not a bad idea, it's just not going to work

        • Re:Not Again... (Score:3, Interesting)

          by rde ( 17364 )
          Why not the computer then?

          Because people are lazy. Given the choice of sitting on their arses and seeing our heroes in glorious high definition is easier than booting, logging on, downloading...

          All this assumes the internet as it is now, of course. For always-on users with speedy access, the computer is as easy. But for the drooling masses television is the first port of call, and if it's available on TV it's not needed elsewhere.

          Also: in the above example, a lengthy search wouldn't be necessary. Interactive->Entertainment->Hot Love Story. Three presses.

          Producers have spent decades making TV the exact opposite of interactive, and they've done an admirable job of it. I know of no better way to kill your mind than TV, even drugs don't do the job as well or as easily.
          QED.
          • That is an excellent point, that people are too lazy to go to the computer and would just stick to the TV. But I think that no one is going to pay for a service like this because it's more limiting than the computer. AOL makes things pretty easy for most people, and provides things like email and the ability for your kid to run Morphius. TV doesn't do that, although they've tried. WebTV was easy as hell, but it's gone exactly nowhere. People are used to the computer, and it's easy enough for them.
    • A few years ago during the web boom (remember that? It was like the CB craze of '77), Veritas was making the same claim.

      They had Clinton's "Monica" testimony indexed so you can search for words (think "cigar") and get to the portion of the video that mentions the words.

      Since most television is closed-captions these days, it's not hard to get searchable text that corresponds to video, once the video is put on some random-access storage medium.

      I really don't think there are any new breakthroughs here; it's just that storage got dirt cheap, video codecs got faster, making it more practical.

  • This would be kinda cool. (Score:4, Interesting)

    by GuNgA-DiN ( 17556 ) on Saturday January 05, 2002 @02:39AM (#2789851)
    Since the advent of the Web, I find myself wishing more and more physical media was indexed and searchable.

    Ever read "Fellowship of the Ring" and wish you could search the book you're holding? Or watched a bunch of shows end wish you could grep for something you remembered hearing?

    As the TV/Computer/Film merge and become more dense we need better ways to pick out pertinent information 'nuggets'. Otherwise, it is just information overload.

    • Sure we all say this sounds great, those who are on topic, but are we really going to allow something so powerful and some what useful become real. The net grew from the computer as a need of sorts. Are we going to allow our behaviour to change for a searchable TV instead of develop?

      The internet has put more computers in homes than Microsoft could ever take credit for, but the TV has been there far to long. As they say, you can't teach an old dog new tricks. It would end being a high priced electronic toy used rarely day to day, a nitch market at best. Just like other "web enabled" appliances. Such a shame.

    • Actually, now you can search Fellowship of the Ring. You just have to download an illicit copy via Morpheus or Gnutella or whatever. :)

    • Since the advent of the Web, I find myself wishing more and more physical media was indexed and searchable.

      Ever read "Fellowship of the Ring" and wish you could search the book you're holding?

      Yea, all the time, and so I eventually came out with a truely ingenious way to do this for non-interactive print media. First we go through all the words in the book and find out what pages they appear on then make a list of these words with their corresponding pages. Of course some words (like "the") appear to often so should be dropped from the list. What your left with is a list of the words most likely to be searched for (names, locations, times, etc.) and where they appear.

      Next - this is really ingenius - we arrange these words in an arbitrary order that is the same for all books. Unfortunately we would have to force everyone to learn this order. I was thinking we should make it really confusing just because we can, like say a vowel followed by three consonants, then another vowel, and so on .... for example: A B C D E F ..... Personally I think this isn't asking too much of people.

      Then we take this index and include it in some convient place in the book. Say, at the back. We could even call it "the index". It's so simple, yet so powerful, I wonder why nobody has ever thought of this before. Hey, I should patent this idea ....

      • Yeah?

        And when is the last time you read a Tolkein book with an index in the back?

        Works great for the Perl Cookbook.... but, non-fiction doesn't tend to use indexing too much.

        [pretty funny post though!]

        • Actually, my copies of LOTR has an index in the back of 'The Return of the King' (not the other two though).

          /me gets up to look at edition information

          "This Paperback edition 1993"

          "Printed in Great Britian by HaperCollinsManufacturing Glosgow"

  • Woohoo!!! (Score:5, Funny)

    by istartedi ( 132515 ) on Saturday January 05, 2002 @02:40AM (#2789853) Journal

    Great. Now we can type stuff into the TV, get 500,000 irrelevant results, get distracted by an I Love Lucy rerun, and wake up 4 hours later trying to remember what we were looking for. Thanks again technology.

    • You make a good point - without proper knowledge management software, or even metadata that helps sort the video into broad stroke categories (ie: "world news", "financial news", "sitcom"), search results would be marred with irrelevancy.

  • Pipe dream (Score:4, Interesting)

    by Supa Mentat ( 415750 ) on Saturday January 05, 2002 @02:41AM (#2789854)
    Ok, I've got well over three hundred channels, a friend of mine has over eight hundred, all of these are constantly putting out new content simultaneously. I cannot begin to imagine the resources that it would take to record this all and then storing it, say digitally, would be a storing all the data that _three_ atom smashers pump out (a shitload of information, and an exageration on my part). There's also the issue of intellectual property, they're gonna have to get more licenses than I want to even begin counting. This seems like an incredibly naieve (sp?) dream. PS first _real_ post. I had to say it I'm sorry.
    • Dunno, maybe get a client software that scans stuff from your local sattelite, cable whatever, then shares the result with everyone via P2P.

      Collect up all the data on the shows showing in the next week planetwide that have clients installed, poll what's important among the clients and distribute the tasks to them from the super-nodes...

      Under that kind of system, there wouldn't be any more live TV. Just a massive pulsing P2P interactive TV experience.

    • Re:Pipe dream (Score:3, Insightful)

      by gnovos ( 447128 )
      Ok, I've got well over three hundred channels, a friend of mine has over eight hundred, all of these are constantly putting out new content simultaneously.

      Technically, it's already stored... Even the livest of live TV eventually goes into the archives at every TV stattion in the world. All that needs to be done is turn those tape archives into digital archives and connect them together.
      • Does nobody else remember This

        http://slashdot.org/article.pl?sid=01/10/27/2033 20 5&mode=thread

        AskSlashdot posting?

        Results: Inconclusive. More likely impossible. Unless you like the idea of 1/4th resolution Greyscale 4FPS video.

    • Re:Pipe dream (Score:2, Interesting)

      by limber ( 545551 )
      hmm. I don't think the resources are 'unimaginable', just ridiculously, ridiculously big. A (really) rough back of the envelope type of calculation on what storage resources might be required:

      - Let's say a DVD holds 8 hours of acceptable quality video.
      - this translates to 15.9 gigs I think? (someone please correct me on the specs if incorrect, I don't even own a DVD player)(what the heck, it'll only be an order of magnitude error or so)
      - TV is what, 75 years old [uwaterloo.ca]?
      - Assume there have been 1000 channels in the history of tv, even though there are doubtless more now, but obviously less in 1950.

      So 3 DVDs or 48gigs = 1 day of storage for 1 channel.

      48gigs* 365 * 75 * 1000
      = (hmm, where's that calculator)

      1,314,000,000 gigs
      =1314000 terabytes
      =1314 petabytes
      =1.3 exabytes

      (I think. Ahem. Please correct any gross errors in translation/calculation. Don't ask me to land the Mars Observer.)

      For comparison, George Gilder notes [lotus.com] that a study showed that the total traffic for the Net was about 1 petabyte a month. (I know, the real figure for comparison would be, 'how many terabytes does the Net contain?' I don't have time for that search! Anyone?)

      Terabyte servers are in use now; it's within the realm of possibility to imagine a million of them. Chuck in the hard drive equivalent of Moore's Law, a breakthrough in holographic techniques, ("Windows 2010 requires a minimum of 1 terabyte of hard disk space...") and hey, you could maybe make some half-assed attempts at this 10 years down the road.

      Of course that's just the storage calculation. I've ignored the gross problems of:
      -digitizing all the historical stuff on tape
      -indexing it all with at least an IMDB-style header of metadata. Full text search would be nice too :-) (hey, a bonus from closed captioning)
      -providing adequate last-mile bandwidth so that 200 million americans can surf different tv programs simultaneously (trust George Gilder!)
      -the IP issues: how hollywood, the courts, and popular sentiment will interact to drive this thing forward.
    • Ok, I've got well over three hundred channels, a friend of mine has over eight hundred, all of these are constantly putting out new content simultaneously. I cannot begin to imagine the resources that it would take to record this all and then storing it, say digitally

      Unimaginable? Not necessarily. Permanently storing all TV produced would be difficult right now, but it wouldn't be impossible to store the last year's programming using today's technology. Assume that you capture the video at 5Mbps MPEG-II:

      Some quick calculations, that's 54 GB of data per day per channel. Sounds like a lot, but you have to realize that such an application can be massively distributed. A single machine with 10 (70-100GB) drives could hold a couple of weeks worth of a single channel.

      Multiply that by the actual number of broadcast channels out there generating "new" content (channels that simply rebroadcast movies and older recordings need not be archived in full, and your PPV and Music channels hardly count), a year's worth of programming could be distributed across a thousand machines scattered across the net.

      That's using today's storage technology, which is increasing dramatically in capacity. The advent of HDTV will set things back, but that's a one-time hit. I would imagine that within the next decade and a half we'll be seeing systems designed to do exactly what we're talking about here. The major obstacle is not the technology, it's reticence from the broadcasters who own the content.

    • 800 channels? Geez.

      Time for a new t-shirt:

      "Only 70 channels and loving it."
  • Quoth the article:
    "How do you do a data link between a TV and a laptop?," Karas asks. "Audio out."

    Look out, folks. It's time for Limey::Convergence! Now all we need is another freebie proctology tool to connect through our keyboards.. just in time, too, my last one just broke.

    Now what would I do with a sig?

  • Nope. Not gonna do it. (Score:4, Insightful)

    by terpia ( 28218 ) on Saturday January 05, 2002 @02:46AM (#2789867) Homepage
    A new startup wants to make all television content archived, indexed, and searchable.

    Won't happen. This could be the cynical conspiracy theorist in me, but do you really think the *media* powers that be will allow even more technology that enables the bypassing of their lifeblood - commercials? Technology that gives consumers MORE control? Media giants have spent the last 2 dozen years bringing the control of what and when you watch to a fine art. Not to mention all the possible copyright and trademark debacles waiting to happen with all lawyers freed up from the death of napster, just waiting for someone to start to bring episode trading to the public's attention.

    My Feature Request for TV
    This is two-fold. First off, I want Satelite or digital cable that changes channels as fast as conventional cable (meaning *instantly*).
    Second part: I want a device that eliminates the stupid and annoying station logos. Contrary to popular belief, many people actually know what damn channel they are watching. Take the TNG episodes running on TNN....how many people need to be reminded that they are watching Star Trek, WHILE they are WATCHING it?

    • Re:Nope. Not gonna do it. (Score:4, Funny)

      by MousePotato ( 124958 ) on Saturday January 05, 2002 @03:14AM (#2789932) Homepage Journal
      Here here. Your requests are right on the mark at least on my list. Additionally features that I want in truly 'Interactive Television' :
      • A button that takes out the mosaics on the Howard Stern Show and E!'s Wild On series.
      • A button that cancels censor bleeps.
      • A goo feature so Neil Cavuto's head can appear even larger on my 60"
      • Something that also gets rid of the obnoxious ticker display all the news channels are constantly running with 'update' information they refresh like twice a day.
      • A compressor to equalize the sound ending thr cycle of volume up / volue down... no more super quiet talk that makes you pump the volume only to rock the cup of coffee out of my neighbors hand when something loud happens at the high level and forces the volume button down again...
      • Infomercial eliminator mode; a temporary killfile for channels that are displaying 'Paid Programming' from the list of available things to watch when scrolling through the channel guide at 3:30 am.
      • A favorite channels version of the channel guide.
      • How about an auto dial or link mode for commercials that take me to an advertisers website?

      Ah... the list could go on...
      • Actually, I'd definately be down for any button to undo all the censors. I just dont think even science can generate a larger head for neil cavuto, and if science could....even an image of an enlarged cavuto head could upset the earth's orbit and send us spiraling into the sun. There's just some things that even science shouldnt mess with.

    • Re:Nope. Not gonna do it. (Score:1, Informative)

      by Anonymous Coward
      This is two-fold. First off, I want Satelite or digital cable that changes channels as fast as conventional cable (meaning *instantly*).

      To do that, the box could request a copy of the currently displayed frame from a server, every time you change the channel. This would require two-way communication (possible on a cable network), and might be a privacy concern (the server would know what channels you're watching - although I hear digital cable boxes already send that info), but would probably give you an image within 50 ms. Another way would be to constantly transmit low-quality/low-resolution images (probably 10-15 per second, compressed using a lossy codec). You'd at least have some picture as soon as you change the channel, and you'd have full quality video as soon as the next keyframe is received (within a second). According to an earlier Slashdot article, digital channels aren't using all their assigned bandwidth anyway.

      Second part: I want a device that eliminates the stupid and annoying station logos. Contrary to popular belief, many people actually know what damn channel they are watching

      I recently used a digital cable box at someone's house. When changing channels, it displayed the channel number, channel name, and (on some channels) the name of the current show. This makes the watermark pretty pointless. If a digital TV company was using one of the methods I mentioned above, they could transmit a watermark image as well. The box could provide an option allowing you to show the watermark always, never, or for the first few seconds after tuning in (after which point it would fade out).

    • I want Satelite or digital cable that changes channels as fast as conventional cable (meaning *instantly*).

      I timed my digital cable from COX. It takes over a second to change channels, and it won't let you switch channels for that time. This is even with the lower analog channels. And I have two different types of receivers that act the same way. Somebody should pull Mr. Cox (or whoever is the CEO) out of his car and savagely beat him! I cannot flip channels anymore, so it totally changes my viewing habits. But getting digital cable is the only way of getting HBO, Cinemax, etc.

      Of course, COX is evil anyways. They kept playing commercials saying how stealing cable is stealing from the community (presumably because they give all their profits to the community or something (yeah right!)). And they made a website so you can report your friends who steal cable. Sure, I understand why they want to FUD everybody into buying cable from them, but sheesh!, this tactic sounds too much like the informant system in communist Russia. When I move in a few months, I vow never to buy any kind of service from COX ever again!

      And I have seen the direct tv and other satellite systems. They are much snappier than cox digital cable (which totally sucks ass).

  • Who's going to sue who this time? (Score:3, Insightful)

    by prof187 ( 235849 ) on Saturday January 05, 2002 @02:49AM (#2789872) Homepage
    I wonder who gets first dibs on trying to slap on a copyright infringement suit of some sort on this idea? I know it isn't really the same as anything, but I'm sure that some company can claim a copyright on it somehow.

    I also wonder how long it takes before someone figures out how to set up a computer based version of the TV so you can stream things to your comp. That might be a good project to start on.
    • I also wonder how long it takes before someone figures out how to set up a computer based version of the TV so you can stream things to your comp. That might be a good project to start on.

      Well, since anyone with the money to buy a tv-card or video card with a tv tuner can already stream tv to their computer - hey, wait a minute - what was the question?

    • But you're still not going to be able to search it just because you have a TV card in your comp. There is going to have to be some kind of program that can connect to and interpret the data. And unless the actual developers are into the FSF, I doubt that they are going to be too keen on handing out software to replace what they are doing.
      • But you're still not going to be able to search it just because you have a TV card in your comp. There is going to have to be some kind of program that can connect to and interpret the data.

        For the product to be viable in the first place, there would have to be a substantial customer base. They will not establish this base by forcing everyone to buy a new tv. (otherwise youd already hae an HDTV (that even got a few hd channels)) They will not be able to write firmware updates for your tv either. This leaves two options, right? Either yet another set-top box (which of course could be plugged in inline with your cable/tv tuner card) or with a device that the company wnats you to believe is not "another set top box" but likely could be a barrel shaped device that plugs inline between your cable and tv and is used in conjunction with a special remote that allows you to send signals BACK to the cable co via the cable or a phone line to drill down through the data and selcet the keywords you want. yadda yadda yadda. The bottom line is that if this is going to even get enough money to start the service, it will be with an add in piece of hardware, that could easily be added to your computer tv setup. - wait a minute - what was the question?
      • "But you're still not going to be able to search it just because you have a TV card in your comp. There is going to have to be some kind of program that can connect to and interpret the data."

        Any decent file sharing application.

        If its worth watching its on there. :) Granted not streaming on demand, but at least its commerical free. :)
      • I'm more concerned with the fact that I don't personally want to buy anything for it. I'd like to see something that could just grab it and save it into movie format. That's what I was talking about, for the cheap people.

  • Fairness (Score:4, Interesting)

    by Spoh ( 241279 ) on Saturday January 05, 2002 @02:49AM (#2789874)
    If it were implemented, how useful would the tools provided be? This raises the same questions as a Google or an IMDB:

    1) Will the database be open to _all_ content providers, or just big-media?

    2) Will search results be fairly reported, or will they be skewed by paid placement?

    I dug around Dremedia.com looking for answers, but couldn't find anything. Has anyone read anything relevant to this?

    -Tom
    • Spoh,

      It has come to our attention that you are not wearing your conspiracy hat. Please put it on and then critically answer your two questions. Hint: both second possibilities are correct.

      Thanks.

  • Frankly, I just don't see any value in being able to search for key phrases out of old Welcome Back Kotter episodes, The King of Kensington, or Three's Company. There's nothing worth looking for in The Electric Company, Rompus Room, or Friendly Giant. The Mutual of Omaha programs were all staged, and aren't worth searching, and The $64,000 Question was rigged, and isn't worth searching.

    Good lord. Here we are, supposedly at the height of human civilization, and Survivor and Friends are the message we're beaming to the future.

    Two hundred years from now, if humanity should last that long, I hope against hope that our descendents look back at us and think "What a bunch of fookin' retards!"

    God help us if they look back and wish for the good old days...
    • Silly slashdotter. These are exactly the garbage that you can ignore once you have a decent search engine! Precisely becuase you can skip over them, that you can get to the things you want to look for!


      You can't pan for gold using your bare hands!

    • I can still remember really cool stories on the news from when I was a little kid, I'd like to look some of that stuff up. Survivor and Friends aren't the only thing on television. Believe it or not theres some really good stuff on something. A database that complete would be a great resource. You want to know about the mating habits of the Great Panda? I bet Discovery or some other network has done a story on that. The list of worthless TV that you've got is pretty impressive, not that I've seen any of it. If this stuff bores you than you shouldn't watch it. This would be a great tool for research and recreation, too bad it isn't at all feasible.

  • This could be really exciting! (Score:4, Insightful)

    by farrellj ( 563 ) on Saturday January 05, 2002 @02:51AM (#2789880) Homepage Journal
    Serachable TV would be great!

    IBM used to have a technology that would allow you to search graphics, call the Ultimedia extentions to DB2. This would allow you to look for, say a red ball beside a tree...and it would return all the images that have a red ball beside a tree...phenomenal tech, but I don't think it was much used. Maybe ths is an extension to that tech, but idexes all the keyframes of show, then putting it into a huge database...

    I would be nice to be able to say "find all Star Trek episodes that show pictures of older ships named enterprise"...

    ttyl
    Farrell

  • Others working on this. (Score:5, Informative)

    by instinctdesign ( 534196 ) on Saturday January 05, 2002 @03:11AM (#2789926) Homepage
    Dremedia isn't the only one working on this. despite the Business 2.0 article's nearly sole focus on that particular company. A few others in the field include, and of course is not limited to, MediaSite [mediasite.com] (which looks to have recently been acquired by the audio and video editing software company Sonic Foundry [sonicfoundry.com]), Virage [virage.com], Pictron [pictron.com] and Vodium [vodium.com]. Its worth checking out each of the sites respective products page to see how they each are approaching this this new field.
    • I'm sure a recently graduated student from Imperial College, UK worked on this as a project. Rather than using speech recognition which consumes vast amounts of computing power and probably storage, it took the much easier approach of indexing subtitles to programmes. And since you'd probably be most interested in searching news reports which are almost always subtitled, you end up with a excellent search system without the flaws of speech recognition
    • I saw an interesting demo on a similar subject at Tsinghua university [tsinghua.edu.cn] and as I remember correctly, they were working together with PacketVideo [packetvideo.com]. According to a Wired article a few months ago, PacketVideo is working on mobile MPEG, but the research that was being done, is easily applied to media searching technology. For example, they were able to search for similar kinds of movements, like the scores in a basketball game. And also for general camera movements, like pan and zoom actions.

      Very interesting!
  • People spend far too much time in front of the TV as it is. The average student spends 40 hours, a full time job, a week watching TV.
    I may spend 40 hours a week in front of the computer but there are things I get accomplished like communicating with distant homosapians and working on what will one day be great pieces of litterateur, or not.
    So what are the long-term results of this service? Will it bring people to their TV looking for a documentary on Mesopotamia? No. It will bring them to look for more useless crap to zone out on. Just like the web gives you the ability to filter out opposing points of view and leave you in an stimulation vacuum of like minded people this tech will make TV a further black hole of non-information.

    As an experiment for all of you who have 100+ channels: unplug the cable/satellite for a week so that you are reduced to just your local stations. Notice that you will watch the same amount of TV, or even more, because you will continue to flip through the channels until you do find "something to watch".

    Save your self now and just unplug it. Life is much better when you do that. I just finished the Autobiography of Ben Franklin, which I would have never seen on TV. And because it takes considerably longer to read than it does to view you get the added benefit of absorbing far more.

  • An archive exists, but only the catalog is online (Score:4, Informative)

    by Animats ( 122034 ) on Saturday January 05, 2002 @03:23AM (#2789951) Homepage
    The Museum of Broadcast Communications [museum.tv] in Chicago has a VHS archive of all major Chicago TV broadcasts for the last decade or so, plus other material from earlier periods.

    News broadcasts are keyword-indexed. Some indexing is based on closed-caption data. Other stuff is just listed by title and date.

    Anyone can view the video, but you have to go to Chicago. It's fun; I've been there.

  • Don't like a series?? DDOS the media server. I can think of a few series that would be worth the jail time to deny their access to the world.
  • How many people would really want to do this?

    Most people sit in front of a television in order to vegetate with Trading Spaces or Everybody Loves Raymond.

    How many of those people would enjoy clicking, typing, scrolling, reading, clicking some more, waiting for pages to get served up, 404s...all while sitting in a Barcalounger that really isn't suited for mousing and typing.

    Maybe the public would like to be able to search for that MASH episode where Hawkeye ad-libs the whole show every once in a while,but I don't think it would happen very often.
  • Commercials on demand..
  • Beyond 2000, that steady beacon of accurate predictions and foretellings said that I would have this from TCI cable by the year 2000.

    Man I can't wait until the year 2000 comes, we'll have so much nifty stuff by then!

  • Is this stuff really new? (Score:4, Informative)

    by chiku ( 471300 ) on Saturday January 05, 2002 @03:49AM (#2790004)

    This stuff is being reported as a very novel stuff. But there has significant research being done in academia.

    Stony Brook (SUNY) ECSL [sunysb.edu] has developed a Videoserver prototype. The difference between this technology and that of ECSL's is that, ECSL videoserver uses closed captions available in the news clips. This way the burden of speech recongnition is taken off the archiving and indexing servers.

    You can read all about it at this page [sunysb.edu]

    This was developed in 1999. This is a well documented project and publicly available. During its initial days it was made available at several download sites. This is still available (documentation + sources) from ecsl website. The only problem is that, this was developed on redhat 5.2 version and used many Beta Stage libraries of gtk(--) etc. Which are now obsolete. It will take a little bit of effort to get it working on latest platforms.
    -- Srikant

    • Dremedia's solution is better and will result in a less costlier solution, compared to what is made by ECSL. The ECSL protoype is more of a "retrievel only" tech, rather than a "indexing + retrieval" type. For one thing, the latter problem is an order of magnitude harder than retrieval alone. For another, you need an already captioned content to start with.

  • There is another [shazamentertainment.com] company that plans to use pattern recognition for recognition of advertisements and music.
    BTW, it is Dremedia's technology which goes into the guts of Autonomy's tools, not the otherway round, as mentioned in the article.

  • This could be a good thing, like being able to find that episode of SNL where Burt Renolds changed his name to Turd Fegurson that you've always wanted to see again. Next thing you know, they'll come out with copy-protected tv's, you'll be SSH'ing into you're t.v., and big brother will be watching what shows you watch just to make sure you're not a terrorist.....jeez I love this country....
  • Page Two of the article is a lot more interesting to me than the first page -- the author describes the technology they're working on RIGHT NOW, and how it might be useful:

    So for now Karas is trying to get his technology embedded in high-end, professional digital video-editing equipment, where it would be used to simplify the tedious process of fast-forwarding, stopping, and scanning for the beginnings and endings of scenes. Instead, video editors would be able to enter specific lines from the script and skip straight to the parts they want.


    This, to me, is real futuristic progress! Forget watching the news at home and clicking on a link to read a related story (who can read an article and listen to TV news at the same time anyway? and if you can, do you want to?). Instead, this sort of technology would let the computer remove tedious, machine-like work for creative professionals -- like searching through 18 hours of footage to find "the bit where he mentions the monkey and laughs" -- and lets the humans focus on the things they still do way better than any computer.

    That's pretty cool.

  • not exactly news (Score:3, Interesting)

    by markj02 ( 544487 ) on Saturday January 05, 2002 @04:43AM (#2790071)
    This kind of work has been going on at CMU, IBM, MIT, and several startups (including Virage) for a number of years. For broadcast TV, you don't even need speech recognition, you can simply use the closed captioning and programming information.

    I don't see this being a big hit with home users: the whole point about TV is to be able to lean back and enjoy the show without fiddling around. Finding and arranging video clips is a lot of effort. People who want to jump around and interact are better served with a combination of text, images, and links to video clips, like what you find on today's news sites.

    • Guess what -- only US TV is closed-captioned, and for a lot of interesting stuff (say, breaking news on CNN) the closed-captioning BITES. Even when it's good, it's not an "accurate" transcript. Speech Rec can get closer, even with errors. And it's essential if you're not interested in US TV.
      • Guess what--you are completely wrong. European television, of course, has closed captioning (it's transmitted differently, but it's the same effect) for the hearing impaired. I believe several Asian countries do as well. And you can bet that CC gets the keywords right. Furthermore, CC gets every speaker in every scene. Even if CC is somewhat of a a shorthand, that's a lot better than speech recognition, which misses entire scenes.
        • Funny, last time I watched BBC (a few days ago) there was no closed captioning. And AFIAK, very few Chinese (CCTV) or Japanese (NHK) stations have it.

          As far as CC getting keywords right -- have you LOOKED at closed captioning lately? You need to be careful -- something like CNN's captioning will be very good if a story has been repeated in their news cycle for several hours. As the news is repeated, the closed captioning is improved via an editing process, since the news doesn't change much. But on breaking stories, or anything the transcriber hasn't seen before -- it's decidedly worse. Also, watch something like Firing Line or a sportscast for a much better view of realtime closed captioning.

          And, more to the point, generating closed captioning is VASTLY more expensive both in dollar cost and labor. Speech recognition currently attains 90%+ accuracy on a problem like broadcast news, for a completely trivial cost compared to human transcribers (i.e. buy a machine, plug it in, it works forever).
          • Funny, last time I watched BBC (a few days ago) there was no closed captioning.

            If you are watching European television on a US set, you won't get it. CC is some limited, oddball US hack. Europe uses Teletext, which provides not only closed captioning, but also news, weather, subtitles, and program information.

            As far as CC getting keywords right -- have you LOOKED at closed captioning lately?

            Sure. Looks fine to me. Speech recognition systems are actually being trained with it.

            Speech recognition currently attains 90%+ accuracy on a problem like broadcast news,

            I think that's a bit optimistic. In any case, my point was not that people shouldn't use speech recognition, but that it's one of many techniques, and that all of them have been worked on by many groups for years. The real problem with these kinds of systems is that they are less useful than you might think.

            • Yes, I'm watching BBC on a US TV. I'm probably behind the times on that.

              I don't know of a single speech recognition system that is trained on CC. I work in the industry -- the transcripts aren't accurate enough. Every piece of training data is generated by a human annotator specifically generating the data for speech rec training. The words are aligned to millisecond intervals. You can't train a speech rec system on a paraphrase.

              Also, the published results of numbers of companies are in the 90% range. These are ajudicated results -- not marketing material.

              I agree with you that speech rec may be one technique -- but it's the only one that scales. People aren't going to be willing to pay to transcribe everything -- they just aren't. But people want (and should be able to) search everything -- form today's CNN to today's TVLand TV repeat. The only practical way to do that is some automatic approach.
              • Yes, I'm watching BBC on a US TV. I'm probably behind the times on that.

                Quite. The system has been in place in Europe for at least 20 years, probably closer to 30 years. It may actually predate closed captioning in the US, which I believe was deployed in 1980.

                Also, the published results of numbers of companies are in the 90% range. These are ajudicated results -- not marketing material.

                Yes, but they are results on "standard databases". Unfortunately, those do not represent real-world performance of deployed systems, no matter how carefully the databases are designed.

                I agree with you that speech rec may be one technique -- but it's the only one that scales.

                Of course, that presumes that the task (indexing broadcast video) is worth doing in the first place, which I am not convinced it is. In almost other video indexing tasks, the video clips come with plenty of metadata. And the only reason that isn't included in broadcast video is because broadcasters have no incentive for doing it (why make copying any easier than it already is).

                • Re: standard databases. The performance DOES actually represent a real-world system, as long as you have reasonable expectations of the real world system. I.e. do not train your system on a broadcast news corpus and then feed it Jerry Springer :-). But if your test and training is reasonably matched, your performance matches as well. There are real-world systems that do this.

                  As to whether the task is worth doing -- well, that's for neither of us to answer, but the market + Virage, et al -- right? :-)
                  • The performance DOES actually represent a real-world system, as long as you have reasonable expectations of the real world system. I.e. do not train your system on a broadcast news corpus and then feed it Jerry Springer :-).

                    Unfortunately, no matter how carefully you design your database in 2001, you can't predict how people speak or how they are recorded in 2003. And current speech recognition systems do not adapt well enough automatically.

                    As to whether the task is worth doing -- well, that's for neither of us to answer, but the market + Virage, et al -- right? :-)

                    The market has spoken loud and clear so far, which is why this sort of thing is not a major line of business for any of the players that started it.

  • Let people watch what they want when they want? They'll never allow it.

  • If you haven't read the article yet... (Score:5, Informative)

    by Alsee ( 515537 ) on Saturday January 05, 2002 @04:58AM (#2790085) Homepage
    Don't bother.

    I'll just fast forward you to the last 2 sentences.

    Now we simply need all the other pieces of the interactive TV puzzle to fall into place. Don't hold your breath.

    -
    • Add a TV card with closed caption capture to your Linux box. Then setup a schedule to capture the text from shows that you might be interested in(use cron anyone). Then use that to build the search data base. Search with grep ... Now the tricky part. Use the same schedule to record things on your ReplayTV/TiVo. How? Well to make it easy use a computer controled IR remote, such as a RedRat2 on Linux [orbit2orbit.com]. Then when you get search hit's lookup the database of shows you recorded and have a macro to jump you to the show and the appropriate offset. A ReplayTV can jumpto specific offsets in a show ...

      To find shows to record you can use something like the power search utility that I wrote for my ReplayTV -> ReplayTV Search Tools for Linux [orbit2orbit.com].

      If your really up for some fun you could get a number of PVR's and play them in concert. The new ReplayTV's (4k's) can transfer shows via the netowrk/lan. Some hackers are now pulling shows off to the PC and viewing them there as well! Archiving to CD/DVD!!!! I suppose you could do the jumpto right on the PC as well, if you have the 4k.
  • Joe Startup: Why is our <troll>beowulf cluster</troll> ripping your monty python episodes to divx?

    Jack Startup: Umm... yes... I'm making TV archived and searchable... or something.

    John VC: Sounds good, have some money.

    Joe S: Where did you come from?

  • Finally we can watch a series in the right order! (Score:3, Interesting)

    by Mandelbrute ( 308591 ) on Saturday January 05, 2002 @05:34AM (#2790128)
    A lot of Australian TV stations show various programs (eg. B5, Jonathon Creek Mysteries) in the wrong order - or several hours after the correct time to allow more airtime of male footballers dressing up in womens clothing.

    It's a bit disconcerting to watch one episode of a program where the two main characters are sleeping together, and then watch another the next week where they are complete strangers that don't meet until the end of the episode.

  • How long till they have an audio and a video tab? One that you can hum a little tune into? or that you can describe a scene and it gives you an mpeg snippet?

    oohhh baby.

  • has anyone else seen these before? they just seemed to pop up for me... oh well -1 offtopic, its cool
  • It aint nothin unless it runs on Linux, My tivo has only crashed no more than 6 times in the last year, I would hate for my PVR to be worse than that...
  • This reeks of cheap ploy to trick investors - 1) A company that has never developed a working product but made great claims to how they are going to revolutionize websearches buys a small company no-one has heard of. 2) An announcement is made to the financial world that this small company has a revolutionary technology that combined with the revolutionary technology of the original company will make everyone's dreams come true. 3) The stock price of the first company goes up. 4) Everyone who knows anything sells their shares and leave, leaving the investors high and dry...
  • Interactive Television Services are only as good as the people who put them together. Not the technology platform.

    When computers learn to be creative then it could get interesting....!

    I think.
  • When I watch TV (not all too often) it's because I have time to kill. I turn to Comedy Central, or some other chanel, and I watch TV till my brain rots. That's all I like TV for. When I'm looking for something SPECIFIC, that's when I visit my good buddy google.com, or a p2p app, like morpheus. The whole point of TV is not to be a searchable database of human history. That's the internet. TV is the thing with all the ads, and mind numbing video, that requires NO USER INTERACTION.
    In short....I like TV the way it is. Use this tech to make web searches work better, not to make TV more work.
  • Even the much more simple concept of digital television is not yet working. MPEG-stream bugging up, unsynchronized subtitles (sometimes several seconds too early/late), program information is more often not available than available, lousy performance, random digibox crashes...

    So how could this work? How much disk space it would need? How about the performance? I don't want to wait while browsing channels. Reliability? 404's in TV is not really something I'm looking for. Accessibility? 500 000 search results are not the answer. Et cetera.
  • Isn't this what the modern P2P networks are? Already on FastTrack Morpheus/KaZaA and Gnutella you can get several of the top 10 movies in DivX format, plus a slew of shows like Star Trek, Sceinfield, even Survivor reruns. I see it as network bandwidth to users increases and processors are able to zip through DivX encoding, this will be the one online, searchable audio (audio is already there) and video technology.

    Don't bother with creating a new network, it's already there and is community supported -- both in infrastructure and in media.
    • how long will it be before any of these get shut down by the RIAA and DMCA?
      using these P2P networks to get tv material will become more controversial as these bastard will start arguing over copyright laws..again.

      if the do create a new network; id be pissed if i had to pay more for using it or be forced to use it.
  • I would really like to see a Teletext-like system deployed in the USA and in NTSC television sets.

    My fiancee' has no internet connectivity due to poor quality telephone lines and ISPs in her country.. but she has Teletext and for many things, it is enough.
  • Now, when they say audio/video searching, how do they mean? How is the inquiry specified? Let me guess: text?


    What I want to see is an A/V search engine that doesnt use text. I want to have A/V samples as my inquiries. Toss in some Cowboy Bebop and Gundam Wing, get Evangelian and some other anime back. No text involved.


    With advancements in AI, this could be possible. Need a picture of an apple for some project or whatever? Give the search engine one picture and get a few thousand back. Does anything approach this sort of technology yet?

  • A wonderful little company based in DC already has this technology working today. Streamsage [streamsage.com] just starting taking its first customers. They're focused on working with groups with large media archives. Imagine searching for information in all the classes at a school or all the meetings in a business. It's a great thing.
  • What about Pay-Per-View.... but with a searchable index of movies.... even if you just put them on a machine in divx format.... and streamed them to the set top box thingy.... just think of how many trips to the video store that would stop :).
    That's just my take on it..... screw the news and
    whatnot.... don't make it indexed .... just make
    a PPV Movie store type thing :).

    L8rs
  • Since Autonomy's software is based on statistical relevance, the speech-recognition technology doesn't have to understand every single word -- just enough to conceptually grasp what a scene or segment is about. The software begins by comparing phonemes -- slices of sound, like the syllables of a word -- to a huge probability table that predicts the context in which those phonemes may appear, and then doing that again with complete words. This language model takes up a lot of computing space and power, and only recently has it become practical to run the software on anything but a powerful Sun server. "Now we can use a twin-processor Compaq with one gigabyte of memory," Karas says.

    What I find interesting is the underlying technology they are supposed to use. Linguists (people who study language, not government translators) have been trying for years to find a way to automatically transcribe audio data for linguistic research, but they haven't had much success. Now they're telling us that a single company is able to do this with any accuracy?

    But then again, the article says that they just have to have an idea about what the scene is about, so maybe they are leaving out a lot of information. Maybe they are being over-optimistic about just how searchable their ``index'' is going to be. Now what would motivate a company to exaggerate the abilities of their product?

  • Where's Google? (Score:2, Interesting)

    by clambert ( 519009 )
    Based on Google's latest jump into the catalog market, it won't be long before we start hitting tv.google.com to catch up on our favorite shows.

Slashdot Top Deals

"The one charm of marriage is that it makes a life of deception a neccessity." - Oscar Wilde

Close