Follow Slashdot blog updates by subscribing to our blog RSS feed


Forgot your password?

Data Mining And The CIA 107

Brotha Z writes "It seems that the CIA has developed a piece of software labeled "Oasis" that can convert the audio from television and radio broadcasts in to text. This software is stated to be able to determine the sex of the speaker, if the speaker is a different person than the original speaker - and if one of the speakers is named, it will continue to place the name next to the correct speaker from that point on. More information on this multi-faceted piece of software can be found here." Hmmm. Sounds like some nice speech recognition technology ("perfect demo" alert!), but as a taxpayer, something rings badly about it. If they're going to use my money to spy on me, can't they at least open source the code so I can dictate a letter?
This discussion has been archived. No new comments can be posted.

Data Mining And The CIA

Comments Filter:
  • by Anonymous Coward
    Since the software was obviously developed with taxpayer money, is it possible that one could request the source for their voice recognition software under the freedom of information act?
  • by Anonymous Coward
    If this is for TV and radio I will eat my computer... This is so they can monitor/log/filter every PHONE conversation ever placed. When they find one that mentions bomb, assasination, napster, etc, it is placed in a que for further analysis.
  • by Anonymous Coward
    Something rings badly about this sentence too. Something smells poorly or looks strangely, but I'm not sure what. Anyway, I have to go, my doorbell is ringing badly.

    All your grammar are belong to us.

  • by Anonymous Coward
    Why don't the networks use it for Close Captioning? AFAIK, they still have people who sit offsite essentially taking dictation really, really fast. They're pretty good, but not perfect -- let's hear it for machine translation.
  • by Anonymous Coward
    Yo, I got the chronic. Bring five freaks ready to do this.
  • deeznutsclan wrote:

    Besides, counterintelligence is the FBI's job, [not] the CIA's.

    Counterintelligence is both the CIA's and the FBI's job. The CIA has the power to investigate any leaks within its walls, among its staff, etc. It's only after they find evidence that they turn matters over to the FBI, who build the case for prosecution and add to the evidence gathering efforts beyond the CIA's walls.

    As for Ames and Howard, it was shown during congressional hearings that there had been plenty of warnings (i.e. alcoholism, spending beyond the officer's means, etc.) way before they were discovered (or caught in the case of Ames), and evidence of either incompetence or plain sabotage, yet their supervising officers did nothing but file a report that nobody looked at until it was too late.

    Again, the most important part of CI work is the human element, not the fancy gizmos.


  • On the other hand, perhaps people who use phrases such as "Da Bomb" should be prosecuted as severly as terrorists and child pornographers. Along with the "Hella cool," "Whaaazzupp," and "Oh no you di-int" folks. I wouldn't half mind the anti-speech laws congress is so fond of passing if they focussed less on restricting the speech of bomb-makers and copyright violators and more on restricting the speech of people who still use the phrase "You go girl" on a regular basis.
  • The CIA, for thinking that any kind of value this software might have will last very long, or the foreign sources who carelessly pepper their communication with direct references to what they're talking about? Think, "let's not call it 'nuclear terrorism', but 'lemon-scented furniture polish'". Let's see how well the software does then...
  • Hey now, have you ever used CC while the channel you were watching had moderate to poor reception? I.E. no serious sound or picture quality issues but still watchable? It can look a geat deal like what you describe. Of course my standards for TV viewing are low after living 22 years in a dead zone for TV reception in an inner ring suburb of Minneapolis, Minnesota.
  • You have to give the CIA credit for at least on thing here. They are definitely much better that the FBI at naming their systems in a way that doesn't spook people. CARNIVORE?!? Sheesh, how did you expect people to react?


  • > unlike the FBI, the CIA at least knows how to create software with a non-threatening name.

    That's because spy agencies are in the habit of giving everything obfuscating code names. 'Oaisis' is just the code name for a system that is actually named "Horny Balrog on an Amorous Rampage", which is about as threatening as it gets.

    Of course, it works both ways. The CIA code name for 'pillow' is "convulsing victim".

  • > his particular piece of technology from the CIA is incredibly cool and very powerful.

    Yeah, when you spend a cold lonely night at a listening post you have to turn of the sound on your p0rn0 flicks so you can hear the people you are spying on, and this will let the agents follow the dialog in the movies, so they will understand the plot.

  • I hate to say this, but there are several companies out there which have this technology. Check out Virage [], Pictron [], and (for the basic technology of facial recognition used by Virage) Visionics [].

    This is absolutely nothing new.
    Only in America will someone order a
    Big Mac, large fries, and a Diet Coke.
  • I get the impression from the article that the software can't do the naming, but if a human labels a voice once the software will use that label next to that voice thereafter.
  • work it out mary




  • As for multilingual text searching and summarisation, the best technology of its kind known to me is Latent Semantic Analysis - the brain child of Thomas Landauer. It's a fairly recent, but hardly secret or obscure, indexing technique that's gaining ground commercially for data mining applications. It can certainly do the the small number of things being claimed by this article. All the relevant papers are on the web.

    If anyone else is interested, you can get the paper on this here []. Links to various formats on the top right corner. It also has many links to related documents.
  • That system was wholly unimpressive. It was distinguishing among four words, and probably based its results on the length of the sound of the words more than anything else.
    Obfuscated e-mail addresses won't stop sadistic 12-year-old ACs.
  • I have a friend who got from a friend an actual transcript of one of the intercepted conversations that has been transcribed to text. Apparently, they were able to actually intercept a terrorist threat from Osama Bin Laden. I hope this proves once and for all the great value of this technology.

    Transcript follows:

    Captain: What happen?
    Mechanic: Someone set up us the bomb
    Operator: We get signal
    Captain: What!
    Operator: Main screen turn on.
    Captain: It's you!!
    Osama Bin Laden: How are you gentlemen!!
    Osama Bin Laden: All your base are belong to us
    Osama Bin Laden: You are on the way to destruction
    Captain: What you say?
    Osama Bin Laden: You have no chance to survive make your time
    Osama Bin Laden: Ha ha ha ....
    Operator: Captain!!
    Captain: Take off every 'ZIG'!!
    Captain: Move 'ZIG'.
    Captain: For great justice.

    - Twid

  • National security exemption. 'Nuff said.
  • It's useful to know what's being broadcast by foreign powers, and who's being featured. The presence or absence of important figures and so forth can be quite useful data...

    You can't really rely on, say, most American news media for good in-depth international coverage, after all -- especially not our television. It's not like ours are state-run and therefore constrained to present, at length, the party line of a ruling government.
  • I don't understand why they specifically mentioned TV and radio

    Because that's where the CIA gets a majority of their information. As much as you would like to believe that they have thousands of secret agents running around, A good part of intel comes from Signal Intellegence. Ever see TV shows when they walk down the rows of country desks? There are hundreds of TVs tuned into "local" news. Read a Tom Clancy novel, he's really good at describing how the system works.
  • While I do admit to having a piehole, I did actually RTFA. What you might consider doing is CTFA, where C = comprehend.

    There. As I got modded down for flamebait in the previous message, I figure I'm obligated to actually post some.

  • Why is it that no matter what the topic, somebody always says "This is old news," as if new technology should simply spring into existence, fully formed and functional. Of course it's based on previous work. All science is. "Shoulders of giants," and all that.

  • The Text Data Mining tool extracted and indexed all words in the data so for example if an analyst was asked whether Iraq ever used anthrax as a weapon, the analyst could open the tool and find anthrax in the automatically generated index.

    I fear that the transcriptions will be tampered with, and noone will double check the 'source code'. I mean, if I wanted a transcription of, say, Kevin Mitnick, I supose I'd get a buncha "BLEEP"s. It's scarry in a way, because we will depend on the data more, and double check less. It happens all the time. A slipped digit here, and granny is on the 10:00 news.
    I don't find it bad that it can track conversations, and recognise voices, but if eschelon were to pipe this on a foreign server, they could send back the results to Intel(the real one duh!). One slight problem, with the eschelon system-the French accused industrial espionage due to losing an airline contract under questionable circumstances.

    What I demand to know, "Who the hell is watching the watchers, eh?" What are their limitations(as per constitution, illegal search and seizure, confiscation of intellectual property, etc.)...
    Anyone who gives up freedom for security deserves neither.
    Ben Franklin

  • Ummm if you don't do anything illegal?? That's making the assumption that all laws are just. We is certainly not the case now, let alone the etreme case where the government gets out of had and needs to be altered. Sometimes in order for progress to me made you have to break laws. We wouldn't (those of us that are that is) be living in the U.S. right now if the colonists hadn't decided to break the law.
  • I remember somebody once saying that the only difference between a highly classified intelligence report and the story about the same subject or event on CNN is that the intelligence report would have more names and more correlating data. Sounds plausible.

    News for geeks in Austin: []
  • I suggest dicing and boiling.

    Nah, pan frying is better. Have you ever noticed how close paranioa and meglomania are? I do almost everytime I read /.

    Kahuna Burger

  • The colonists broke the law to form the US. I live in Canada. We managed to become our own country without breaking the law. We just followed what the population wanted and used to proper channels to get what we wanted.

    ah, but in the US, the "patriots" weren't willing to actually wait until the majority of the population agreed with them. Between the loyalists and the apathetic, revolution was not a majority position. Thats why the revolutionaries had to use propaganda (such as the yellow journalism surrounding the so called "boston massacre" that we just celebrated the aniversery of) and lies to whip up enough support to start something, then once they were violent enough to get a retaliation that pushed more of the apathetics into opposing the British "oppressors".

    History is a funny thing. Things that everyone knows turn out not to be true sometimes.

    Kahuna Burger

  • Imagine, if the patriots had lost, then everyone would be celebrating the day the evil seperatists lost.

    well, in fairness, we don't have an "anti confederate day" celebration or anything. But if the colonists had lost, would there be hardcore yankees who called themselves "declarationists" and wanted to use the 13 star flag on license plates? And argue for the right to fly said flag over state capitols in the orriginal colony states? And would officials of the Angican church say that such was just "anti crown bigotry dressed up as cultural pride"?

    Well, we can't take these things too far, I guess. :>

    Kahuna Burger

  • I hope so. You can always tell the automated ("voice rec") closed captioning systems because they read line:


    I think deaf people who rely on CC must have a really warped sense of english. Sounds like the CIA does too, now.

  • but...

    I don't quite like the idea that they have the ability to separate one voice from the next digitally, but I guess that in some distant future all that will be done flawlessly and we'll all be screwed as far as that goes.

    And by the time that happens, they will also have a very firm grasp on digital voice production, meaning so much for the freedom of speech ... or so much against, I should say.

    And open source? Hah! Let's just open source the Pentagon while we're at it... Actually, I wouldn't mind knowing what really happened to Marilyn Monroe...

  • You forget. This is the CIA. Are you trying to tell me that they are not cataloguing? If you are, forgive me, but you're crazy! That's just the part they're not telling you about...
  • That's nothing new. Researches has been done on stuff like this for years. It is just unfortunate that the CIA is going to use these technology now...
  • theory is the operative word
    who's watching the watchers?
  • If the CIA were to sell this software to the NSA, then they (the NSA) would be able to do wacky (read: invasive[pervasive?]) things with cell phone converstions, but they do already, so I'll just shut up now... Eh, just reply with thoughts... /me looks foreward...
  • "This software is stated to be able to determine the sex of the speaker..."

    Bart Simpson: Male
    Bart's Voice: Female (Nancy Cartwright)
    Software says: Shemale?
  • The sad thing is Information is so valuable. the would never open up a database like this, I drowl at the thought but can't help think that this is 10 years off.


  • there are such alarm clocks, i have one. of course it snoozes with any noise, as long
    as the sensor picks it up (and there's two sensitivity settings for it).
  • I don't know about you, but I have a hard enough time finding anything in the 200 gigs on my system. Signal to noise, signal to noise.. Scanning tv & radio sounds like more noise than signal to me.. eventually *everyone* will have the oppourtunity to be employed by the CIA to sort through this data..

  • Most speech recognision systems use some form of a dictionary and grammar rules to help the accuracy of the translation. Could a language like pig latin throw the system off?

    There must be many ways around this technology (encrypted voice over ip comes to mind) but the only ones who will take the time to do so are the supposed targets.
  • I guess it is not as simple as that. Yes, we
    do contribute our tax money towards it. But if
    it is open sourced, don't you think we are
    throwing hard earned technology to enemy
    spies for nothing too?
  • You got it the wrong way around, Taco. The CIA spy's on other countries. It's the FBI that spy's on you, and the NSA that breaks your encryption.

    Anyway, I work for a defense contractor on classified projects. The whole point of having the project classified is so that the other countries don't know our secrets. We do missile jammmers. It becomes pretty pointless if we open source the code/firmware so you can jam the police radar, because then other countries know how our jammer works, and simply make their missiles act in a way that our jammers can't stop.

    The same may go for any software the CIA designs. If they explain thoroughly to you how they're spying on other countries, then the other countries know how they are being spied on, and they take actions to prevent it.

  • Read this in the document:
    Languages that FLUENT can translate into English include Chinese, Korean, Portuguese, Russian, Serbo-Croatian and Ukrainian.
    This is extremely funny, the so-called Serbo-Croatian does not even exist...
  • Does this mean I have to stop saying "Someone set us up the bomb" and "You have no chance to survive make your time" when I'm on the phone with my friends from now on?

    "Everything that can be invented has been invented."

  • FLUENT sounds like an excellent tool. Imagine how much more meaningful internet search tools would be if you could search non-English pages. Does anyone know if there is any search engine that can currently do anything similar (even if inferior)?
  • So uh.. does archiving all broadcast transcripts fall under 'fair use'? I'm not sure it's really time-shifting.

    Let's just hope the RIAA doesn't hear about this!

  • Everyone's paranoid that the government is out to spy on them. Look, if you don't do anything illegal, what do you have to worry about? What's the alternative, a wonderful government that doesn't spy on anyone internal or external? Hey that'd be great, but wouldn't work unless every other government was the same. Then one would spy on the other and BAM!

    On the other hand something like this would be great with some controls in place as to how the government uses it. But I don't believe for an instant that such a device works as well as claimed. Maybe someday in the future, but not today.

  • Imagine, if the patriots had lost, then everyone would be celebrating the day the evil seperatists lost.
  • But if you live in a democracy, then the laws are made by the majority. What is just is determined by popular opinion. Back in the times of ancient greeks, petophilia was okay, now it isn't. We can't say they are wrong, just from a different era with different values. To us they are icky and gross.

    The point is, if the majority of the population does not believe something is right, they change. Or they put someone in government that will. That is the way democracy works (at least ideally, I know it's not that simple, but the concept holds true to some degree). If enough people wanted pot legal, you see people that want pot legal being elected into office. Get enough of them and you suddenly have legal pot.

    Warning, my opinions do not come with a warrenty.

  • The colonists broke the law to form the US. I live in Canada. We managed to become our own country without breaking the law. We just followed what the population wanted and used to proper channels to get what we wanted.

  • Ifway Ooglegay ancay indfay ethay acespay otay archiveway ethay internetway, onday'tay youay inkthay ethay IACAY ouldcay indfay enoughway acespay otay archiveway allway ofway esethay oadcastsbray inway ASCIIWAY ormatfay?

    IWAY ersonallypay ouldway OVELAY otay eesay away ugehay earchablesay, onway-inelay atabaseday ofway everythingway everway aidsay ybay anyoneway atthay asway oadcastedbray. Imagineway ethay implicationsway. IWAY'day earchsay orfay allway ofway ymay ocallay oliticianspay otay eesay ifway eythay everway aidsay anythingway upidstay inway eirthay eviouspray ifelay asway away okedcay-outway-Iamimay-elevangelisttay. IWAY'day alsoway earchsay orfay ymay ownway amenay otay eesay ifway IWAY issedmay away ongsay edicationday orway anway NPRAY onsorshipspay inway ymay amenay.

    IWAY uessgay away otablenay awbackdray isway atthay ethay IACAY ouldcay ettypray easilyway anscay ellcay-onephay andwidthsbay asway ellway... ocumentingday anyway 'otablenay' ivatepray onversationscay. Erhapspay eway ouldshay allway artstay alkingtay inway igpay-atinlay otay avoidway ethay IACAY'say attentionway, alway alay Apsternay?

  • ... to determine sex of SlashDot posters by their comments. Anne Marie would be able to prove that she is not Signal 11.
  • I wasn't implying any sort of exotic technology, just saying that, as it doesn't currently run on anything resembling normal hardware, it most likely isn't anywhere really near a point where it would be something that would be used by, say, the CIA.

    I imagine that this *could* eventually be simulated by some sort of software- can't see any reason why not, but, as you mentioned, it would probably require a lot of processing balls to simulate in realtime, which is basically what I was saying.

  • This could be useful in helping the deaf.

    I'm sure that there are a lot of TV programs and older shows/movies which are not closed captioned. This technology could make this type of media more accesible to the hearing deficient.

    Also, if embedded into a cellphone, I could READ radio chats and interviews and stuff when I don't want to create noise for other people or look goofy wearing headphones.

    And of course, militarily, it could be used to monitor every voice communication in the world to search for things that the deployer might want to know about. Eg, planned acts of terrorism, organized crime activity, exessive use of free speech, etc. And you wouldn't need to save it as a bulky audio file.

    O'Toole's Commentary on Murphy's Law:

  • if one of the speakers is named, it will continue to place the name next to the correct speaker from that point on.

    And it converts speech on TV, so what happens if The Terminator plays on Fox and 2 hour later Predator plays on TBS? Does Arnolds character in Predator get named "Terminator"? And suppose a comedian does a decent impression of Arnold, does The Terminator suddenly appear in the transcript of his comedy routine?
    One other thing bothers me about this. The software was developed by the CIA right? Why is the CIA interested in knowing what happens on TV and radio? Well duh, they aren't. I'd bet this is going to be used to tap phones and eliminate the cliche of the agents smoking in the dark room above the bad guy's apartment anxiously listening to his phone conversations. (Or do they already have something for that purpose?)

    "// this is the most hacked, evil, bastardized thing I've ever seen. kjb"

  • There are many technologies that you pay for that you didn't approve of or necessarily reap benefits from - at least not immediately. Eventually, these things get moved out into the open (like tracking code from GPS satelites) whether the inventors want it to or not. This particular piece of technology from the CIA is incredibly cool and very powerful. This alone will make people eager to get it...and it will be gotten. Then all sorts of interesting things will be done. Be patient. Information/technology wants to be free (and open of course). It'll find its way into the mainstream. :-)
  • They've already had technology like this for a while. So do large companies, and it's really expensive stuff.

    Maybe the CIA improved the AI somewhat, but it sounds pretty similar. At least they're telling us more about it, though.

    However, I'm sure slashdot will still link to RealAudio every time someone gives a talk, so I don't care. If they think I have anything incriminating to say, then they're wasting my money.

  • Oasis: brackish pool of filthy disease-ridden sludge surrounded on all sides by a desolate wasteland under an unforgiving midday sun.

    Suitable metaphor for The Company?

    Sometimes nothing is a real cool hand.-- Cool Hand Luke
  • I'll check the 0day sites for it..
  • > State-of-the-art recognizers have an error rate of ~10% on that test

    Well, it is more like 15%, especially if you want to do it fast (this typically means 10 times slower than realtime). One can't really afford to run at 300xRT for large-scale transcription. When we had to recognise 500 hours of broadcast news we ran our system at around 15xRT.

  • What I demand to know, "Who the hell is watching the watchers, eh?" What are their limitations(as per constitution, illegal search and seizure, confiscation of intellectual property, etc.)...

    It's the CIA. In theory they are not allowed to surveil US citizens on US soil. So most of that stuff doesn't apply.

  • Speech recognition, transcipting, and subtitling of audio-video content helps all of us, particularly the deaf, blind and sportsbar drinkers. Unfortunately speech recognition is not perfect []. Good speech recognition could save the CIA a pile on FBIS []. Searching text transcripts of a/v files, is only the start.

    Internet2 [], a gigabit network [] for education and research [] (see PDF map) [], has a major future use as an audio-video storage library and distribution network []. Video-napster? CMU's Internet2 Informedia Library project [] researchers are designing visual-video search software [] for faces, on-screen text, images and shapes. Computers finding on-screen people, text and similar programming... scary.

    Check out this presentation with screen shots about Internet2 [], and its cool tools, uses and experiments. Slide 36 [] shows Facial Recognition and Optical Character Recognition (OCR) at work. It works so well, it finds text (bottom right) on the U.S. Capital's dome columns... whoops. Slide 37 "Similar Shapes/Content" shows examples of similar content of female news anchors [], and soccer / football [].

    remove the nofreakinspam. to e-mail me.
  • unlike the FBI, the CIA at least knows how to create software with a non-threatening name.
  • The US government aren't the only people to have cottoned on to the power of convertnig spoken words into text. The THISL [] project aims to provide broadcasters and other news gathering organisations with a powerful tool using this tchnique.

    Firstly the news archive is passed through a speech recognition system which basically produces transcripts of every news item. Then a powerful text search may be applied to the database to locate information relevent to a particular topic. If this is being done for research it is most likely all the information required can be gleened from the transcript however the original recording may also be retrieved using the archive reference stored in the database.

  • What this article neatly avoids mentioning is that this is the Killer App for the Echelon project. They've been scanning the communications of foreign countries for quite some time. Now they've finally admitted to having to the software to filter through the massive amount of data they acquire. It's only logical, isn't it?

    The article referenced only mentions using it on TV and radio broadcasts. While I'm sure that it was probably developed and tested with TV and radio, it could just as easily be used to monitor cellular and private radio communications. This is, after all, a spy agency that we're talking about.

    Now they don't need to have people whose job is to monitor the cellphone conversations of select suspected/known terrorists. Instead, they simply collect all the data and run it through OASIS. Then do a few keyword searches (using FLUENT if need be) for words like "bomb," "Iraq," "Jihad," or "assasination." They could maybe even do more complex searches (+bomb +embassy). Now that they've got a report, an agent looks over it to determine which conversations are interesting ("Did you get the bomb making supplies?" vs. "Yo! Kobe Bryant is da bomb!!!") and starts investigating.

    Isn't technology wonderful?
  • I know what you mean. On a good day, you get the occasional rearranged word. On a bad day, you stuff that looks like that perl DVD decoder.
  • Maybe the CIA improved the AI somewhat

    I'd like to know more about how accurate it is before I get all upset about it. Yes people have been doing stuff like this for years, but it hasnt been all that robust so far. If the CIA got it to work like the article makes it sounds and with good accuracy then they have some pretty nice software. Otherwise, its just more PR to make the president and congress give them a pat on the back and more money to play with.

    I wouldnt be too suprised if it is really accurate though. The government, at least the part that likes to keep a low profile (CIA, NSA, etc.), always seems to be about 10 years ahead of the public on the technology curve

  • Dragon Dictate or IBM's voice recognition. It better be a whole lot better than off the shelf products considering how much the CIA must have spent developing it.

  • Combining all these technologies offers some awesome possibilities.

    I'd really like to see this kind of technology applied to search engines. My experience on speech recognition and computer translation is rather limited, but I think the available software does needs lots of refinement before it can be applied to things like this. They (CIA, FBI, DEA) are after fool-proof evidence to be used in court, after all.

    An experiment with normal translation software English-RANDOM()-English shows that sayings like 'Oh dear' may convert to likes of 'Butter love'. And this with grammatically correct and polished english. If the language used is less correct, what do you get? Imagine all the errors a typical speech-recognition software does.

    Now, if this is applied in practise:

    1. Tape all cell-phone calls of a suspect terrorist or drug-smuggler. Now, these might be in grammatically less-correct heavy-dialect spanish or arabic or sanskrit or whatever...

    2. Let your speech-recognition software convert this to text.

    3. Translate the text to english with translation software.

    4. Arrest the guy and try to convice the jury he deserves a prison sentence.

    5. Get (literally) laughed out of court.

  • See this article [].
  • Gosh, and I thought OASIS was some software for re-issuing the Beatles music as "new" tracks.
  • do the same thing with phone?
  • Yeah, Serbo-Croatian doesn't exist now; but it used to until 1991 - it was the official language of Yugoslavia. (AKA Serbo-Croat). After the country broke up, the Serbs and Croats decided that since they were obviously so different in other ways, they must be speaking different languages, and invented Serbian and Croatian - two languages that just happen to sound almost identical to each other (and to Serbo-Croat).
    Pedantic, I know, but there you are.
  • by Anonymous Coward
    This kind of technology has been around for quite some time now. The first I heard about it, was a few years ago at a German universitary research institute, the DFKI. For those interested, one of the relevant projects is Olive, and some information can be found at [].

    The bottom line of this kind of technology is that although the speech recognition itself is relatively poor it is helped by the fact that most of the interesting words (names of people, places, etc.) occur very often in the same segment. So, it's all statistics. No accurate transcription needs to be made to achieve this kind of result. Therefore, applications such as automatic sub-titling are not possible with such systems. And I think they are still quite far away too.

    As for the CIA claiming break throughs, well, I think other people can say wittier things about that.


  • : I don't understand why they specifically mentioned TV and radio.

    Large vocabulary (but somehow predictible), speaker trained to overarticulate, no superposition between different speakers, slightly simpler language model (complete phrases, language close to written).

    State-of-the-art recognizers have an error rate of ~10% on that test, which until last year was one of the evaluation tests at the speech group at NIST. Check [] for details.

    Since the point of disminishing returns was reached, the test in going to be replaced with a new one, a audio/video recorded meeting transcription. Much, much harder.

  • I remember a bit in one of the Tom Clancy books in which Jack Ryan tell his wife that CNN often knows stuff faster than the CIA. I belive its probably true at least for some types of things. And I would bet hevily that in places like CIA and the Petagon that TV's tuned to CNN are rather common.
  • Probably a few tens of thousands are what CIA is interested in. I'm sure that at CIA or NSA there is a room of people listening in to the BBC World Service and Radio Moscow, CNN etc. Lets face it the CIA does not have agents anywhere and much of what you can get off of the public wire is good information.

    Plus you can also find out what world leaders are thinking by reading the newspapers in a country and listening to the national radio station. I would imagine that something like a Tivo would make this much easer for them.
  • From the last sentence in the article:

    Another intelligence official, on condition of anonymity, said: "If they have this kind of technology to plumb the depths of open sources, you can imagine what kind of technologies they have to track down spies."

    All this technology wasn't good enough to track down Aldrich Ames, Edward Lee Howard, or the FBI's Hannsen, who together are probably the biggest moles in the history of espionage. People forget that tools are useful/automatic, but they aren't intelligent. Someone must be at the controls to interpret and act on the data. This tool sounds great, and there could be potential civilian uses beyond CI, but people must remember it's only a tool.



  • Also without bothering to RTFA, I'll repeat Paradise_Pete's question: Do you know what a neural net is?

    You see, as I assume was his point, "computer chip neurons" work differently from central processing units, but not from the "data structure neurons" that can be trivially implemented in a program running on a "regular computer" to simulate the exact same neural net. The fact that they did it in hardware is interesting in its own right to someone interested in neural net research (I'll probably go read it later), and perhaps the speed factor is so great that a software version couldn't run in real time (which I guess could be what you meant) or would require an astoundingly powerful and expensive conventional computer in order to do so, but there is nothing special about "computer chip neurons" that in principle prevents the same thing from being done in software on a "regular computer".

    Maybe this truly "doesn't run on regular computers" simply because they haven't implemented such a sofware-based simulator, but that's very different from implying that it's based on some kind of exotic technology that a Von Neumann machine is fundamentally incapable of duplicating, which is what it sounded like you were claiming and which is probably what Paradise_Pete objected to (and was wrongly punished for).

    David Gould
  • Because Closed Captioning also includes contextual information and is often not a literal transcription but a synopsis slighty rewritten to contain the originial elements but shorter as so to be more easily read on a TV screen while remaining synchronous with the action.

    My English teacher would cringe at that run-on.

  • The moral of this story is, if you're a spy, don't televise your meetings with your control.

    I mean, like, really, now, dude. Are they going to start scanning soap operas for the sake of national security? Is Jay Leno broadcasting national secrets? Someone clue me in on the intelligence application here.

    I suppose it might be handy for transcribing the numbers stations, though somehow I doubt that they'll seem quite so glamorous in ASCII:
    12 2 9 78 16 1 289 8 6 89 9...

  • Sadly, I think we have as much chance as seeing Microsoft open the source to Office as we do of seeing the CIA release this.

    Then again, stranger things have happened. But I would bet the proverbial farm that the guts of the software is Classified.

  • The article is not so much about speech recognition (as some other comments have mentioned). It deals with the possibilities of being able to label speakers, storing data of all kinds of sources in a database and being able to detect previous statements of a person. So, this is more about the intelligent combination of various existing techniques (including speech recognition and machine translation).

    Personally, I think this has been done before to a certain degree, the resources available to the CIA (and their counterparts) are just becoming incredibly huge. Given the increasing amount of traffic that is generated by Internet users, they're probably pretty happy about that.

    On the terrorists who are being mentioned all the time in that article: they're probably using encryption technology anyway, so I'm not sure if the really dangerous people will be caught with that system.
  • "In benchmark testing using just a few spoken words"

    And why did the benchmark only involve a few words? Because that's all it can recognize. This thing isn't doesn't do speech recognition, it does sound recognition; IIRC, it can only handle single syllables words, and only four or five at that, and no sound-alikes. (I think "yes" and "no" were half its vocabulary.) It might be breakthrough for such a small ANN, but it's not that useful as a natlang system. I suppose something similar could make a good front-end to more complete system, though.

  • Not a legal vote count sorry. During the election every single republican uttered the same mantra. "Any subsequent counts done by any press will be unreliable and false". You seem to be contradicting howard baker there.
  • I now a large group of paranoid people who like to start all of their unimportant phone conversations with "I'm going to kill the president" or some such giberish because they are firmly convinced that all telephone conversations are being monitored by some Echelon type system, and have been for 20 years. They believe, that by throwing such "Noise" out there, they're helping protect everyone's privacy.

    What amuses the hell out of me though, is that this kind of works against them if their own theories hold true.

    The way I see it, almost nobody else goes to such efforts no matter how paranoid they are, and even if some phone-listening machine was being put to use, all they're doing is ensuring that they will be listened to.

    And it's not that I don't think this sort of thing goes on or anything, it's just that I don't bother fighting it anymore now that they're able to read (and control) all of our minds anyway.

    "Everything you know is wrong. (And stupid.)"
  • Unless it is classified it is supposedly public data, so you should be able to get a copy of the source code through a Freedom of Inforfmation Act request.

    Anyone got a couple of spare lawyers looking for a fun afternoon or twenty?

  • Wanna see the results?

    I suspect the most common use of this sort of software is to monitor foreign broadcasts - something the CIA/OSS has been doing for more than 50 years. Traditionally, this has been done through a group (mentioned in the article) called the Foreign Broadcast Information Service (FBIS). FBIS monitors newspapers/broadcasts of many, many non-US media sources and makes this information available to US Government agencies.

    For many years, FBIS made available to the public a daily paper copy product via the US Dept of Commerce's National Technical Information Service (NTIS []) that was fedex'ed daily to hundreds of subscribers around the country/world. There were several issues, broken down by regions. For many years, it was one of the best public ways to track what was happening in the Soviet space program.

    It's widely known that FBIS/CIA as been developing and using technology to aid the translation process for many years.

    A few years ago, they dropped the paper product and moved to an electronic version.

    The FBIS server to distribute the information to US Government users can be seen at [] and can be found via a simple Google search on "FBIS".

    The public can access this information via NTIS's World News Connection system ( []). Yes, there is a charge to use WNC, because NTIS has to pay copyright (gasp!!!!!) to the foriegn sources (just because you steal the data stream doesn't mean you own it!) as well as operate the system. It's pretty well known that foriegn sources who complain loud enough also get paid by the Govt for the US govt use of the data.

  • My guess is that it's really fairly poor speaker independent stuff. It probably does a quick, low quality word recognition algorithm

    It doesn't always have to be speaker-independent. Since it doesn't have to be real-time, all you need to do is identify the speaker, and then start over. If we're really talking about TV and radio sources, then there are going to be a large number of regularly-appearing speakers. Just a SWAG, but I'll bet that under a million people account for 80% of all the TV and radio minutes worldwide.
  • Here's some fitting OASIS Acronym definitions from the

    Observation At Several Interacting Scales
    Operational Application of Special Intelligence Systems
    Oracle Application Software Implementation Strategy

    "My one oasis in the dust and drouth Of city life."--Tennyson

  • I don't understand why they specifically mentioned TV and radio. If the audio is digitised before being pass to the software, it doesn't really matter where it comes from. Maybe they are trying to draw attention from the fact that it can be used on things like making transcripts of phone calls, normal conversations recorded with various listen devices?

    I don't think you realize how boring and mundane most intelligence work is. Thousands of extremely junior people sit all day long translating newspapers and transcribing radio/TV broadcasts. Much of this stuff is made available through FBIS (pronounced "fibis") to further bore people slightly higher up the ladder throughout the government and contracting agencies.

    However, it is useful once in a while. Especially when looking back and saying "Now how didn't we catch that?" If it could be brought online cheaper and more quickly, I can see how this would be well worth the money - without being particularly draconian (except insofar as the concentration of enough otherwise innocuous information can be quite powerful).

    Sometimes, just sometimes, they mean what they say.

  • If Google can find the space to archive the internet, don't you think the CIA could find enough space to archive all of these broadcasts in ASCII format?

    I personally would LOVE to see a huge searchable, on-line database of everything ever said by anyone that was broadcasted. Imagine the implications. I'd search for all of my local politicians to see if they ever said anything stupid in their previous life as a coked-out-Miami-televangelist. I'd also search for my own name to see if I missed a song dedication or an NPR sponsorship in my name.

    I guess a notable drawback is that the CIA could pretty easily scan cell-phone bandwidths as well... documenting any 'notable' private conversations. Perhaps we should all start talking in pig-latin to avoid the CIA's attention, al la Napster?
  • "If they're going to use my money to spy on me, can't they at least open source the code so I can dictate a letter?"

    I may be wrong, but doesn't the CIA's charter say that they cannot conduct operations on native soil?
  • by eric2hill ( 33085 ) < minus punct> on Wednesday March 07, 2001 @04:48AM (#379724) Homepage
    Just this morning I was joking with my wife about buying an alarm clock that snoozed when I yelled 'shut up', 'piss off', 'go away', or 'it's saturday'... Maybe this technology will lend itself to alarm clocks in the future :)

    Morning sarcasm. I'll get back to work.

  • by Fervent ( 178271 ) on Tuesday March 06, 2001 @08:48PM (#379725)
    I just want the damn TellMe service to work. How many times do I have to say "New Jersey Devils" before the sports program on that thing recognizes "Hey, he's talking about a hockey team!"

    Beyond that, the TellMe service should also recognize the command "shut up" along with "stop" and "tell me more". I mean, if you're going to have a voice-activated phone portal, why not use "natural language" for commands? ("Shut the hell up you stupid bitch! I said "stock quotes" not "stock racing"!)

    For those of you who have no idea what I'm talking about, dial 1-800-555-TELL. The service is free, for now.

  • by dietcrack ( 219911 ) on Wednesday March 07, 2001 @02:18AM (#379726) Homepage
    Speech recognition does exist that, if not 100% accurate, has been demonstrated to be significantly more accurate than human speech recognition. It was being developed at USC while I was there. Something about a neural network of some sort, so it doesn't run on regular computers, but, and I quote,
    In benchmark testing using just a few spoken words, USC's Berger-Liaw Neural Network Speaker Independent Speech Recognition System not only bested all existing computer speech recognition systems but outperformed the keenest human ears.


    The system can distinguished words in vast amounts of random "white" noise -- noise with amplitude 1,000 times the strength of the target auditory signal.

    I don't know about you, but I'm pretty damned impressed.

    the article on this system []

  • by Codeala ( 235477 ) on Tuesday March 06, 2001 @10:01PM (#379727)
    ...that can convert the audio from television and radio broadcasts in to text.

    I don't understand why they specifically mentioned TV and radio. If the audio is digitised before being pass to the software, it doesn't really matter where it comes from. Maybe they are trying to draw attention from the fact that it can be used on things like making transcripts of phone calls, normal conversations recorded with various listen devices?

    About that feature that id the speaker, imagine a conversation that goes like this:

    Speaker 1: You the Man.
    Man: No, YOU the MAN.
    Man: No no, you Da Bomb
    Da Bomb: Hehe

    Watch word: BOMB Alert! Alert!

    As a final side note, I won-der... if... it... works... if... you... talk like... Cap-tain... K-irk... ;-)


  • by vlax ( 1809 ) on Tuesday March 06, 2001 @09:44PM (#379728)
    They don't seem to have very accurate speech recognition technology. The article claims to reduce transcription time by a factor of about nine. That's a lot less unreasonable than believing in good speech recognition technology.

    My guess is that it's really fairly poor speaker independent stuff. It probably does a quick, low quality word recognition algorithm - quite a few of those are around - and then some sort of Bayesian network to correct the transcription using lexical context. I know that ARPA was openly funding people doing exactly that a few years ago, and I'll bet their papers are on the web. It doesn't shock me greatly that someone has had some measure of success with it.

    If it was 100% accurate transcription, then I wouldn't believe it. But as a time saving device for transcribers... that I find credible.

    DARPA also funds a lot of automatic topic spotting research. One of my ex-profs received grants from them under just such a rubric and her papers are publicly available on the web. I'll bet whatever technology they are using, it was developed by a prof at an open university who publishes freely.

    As for multilingual text searching and summarisation, the best technology of its kind known to me is Latent Semantic Analysis - the brain child of Thomas Landauer. It's a fairly recent, but hardly secret or obscure, indexing technique that's gaining ground commercially for data mining applications. It can certainly do the the small number of things being claimed by this article. All the relevant papers are on the web.

    In short, this doesn't sound like super-secret spy stuff. I'll give long odds the real work is in journals and webpages that are publicly available. Having a couple billion dollars to speed up testing and implementation probably helps, but none of this sounds revolutionary or years ahead of the curve.
  • by Animats ( 122034 ) on Tuesday March 06, 2001 @10:02PM (#379729) Homepage
    Much of what the CIA does consists of collecting publicly available information. Some of this they now distribute to the public. The CIA World Factbook [] is the best known example.

    Less well known is their Foreign Broadcast Monitoring Service [], for which generations of linguists have listened to the hype output of governments worldwide. (FBIS refers to this as "open source" material.)

    They've been hoping for years to automate some of this stuff, and apparently they've succeeded. It doesn't require particularly good speech recognition, since the basic goal is to pull out the interesting stuff from the endless drivel.

    This sort of info is used to answer questions like "Is country X changing their policy on Y", and "Who is speaking for country X on subject Y?" This is basic political intelligence information.

  • by rgmoore ( 133276 ) <> on Tuesday March 06, 2001 @09:38PM (#379730) Homepage

    Actually what it sounds like the CIA is working on is trying to mine data out of public sources. There's good reason to think that you can discover a lot of what governments want to keep hidden if you can just go through enough publically available data and correlate it. For instance, you can probably get a good idea of a government's secret spending by figuring out how much money they're taking in taxes and borrowing and subtracting out expenditures- provided that you can actually track both of those things. It looks hopeless because there's so much data to go through, but with good computers it should be possible, especially if the other guys have a lot of secret spending. Or you can figure out what the inner circle of the government really thinks by looking at all of the news leaks from highly placed government officials.

    This stuff scares the crap out of governments that are both required to be open but interested in hiding things from other countries. You simply can't hide everything, especially not anything big enough to be really interesting, because it has to interface with the world somehow. The CIA obviously wants to get really good at this kind of thing, and monitoring vast quantities of mundane stuff like TV news programs, budgets, and corporate annual reports is part of the process. The best part is that if you can do this effectively, you don't need spies as much, but you do need a lot of drones to go through huge piles of paper and TV to enter the raw data into the computers to process. There's probably some filtering out the interesting stuff from listening in on videoconferences, too, but it's amazing how many paper pushing drones wind up working in a sexy sounding business like spying.

The unfacts, did we have them, are too imprecisely few to warrant our certitude.