Ladies and gents, all aboard, the review train is coming. It was a great morning today down at the Speech Tech offices. A huge package arrived from Nuance just stuffed to the brim with copies of NaturallySpeaking 10.
You see, my brother Adam B. and I have been angling to get some review copies for a couple of days so that we could give you fine folks an unbiased and unvarnished look at it, and Nuance has delivered. Unfortunately, we’re probably not going to be able to get cracking on it until the week after next since I’m going to be down San Diego-ways for Voice Search, but keep those eyes peeled for a full spread.
Everyone in the office will be taking a look, charting their experiences with clipboard and air of scientific impartiality. Our editors have even started taking to wearing lab coats around the office (though admittedly that’s unrelated and has more to do with our having turned the conference room into a meat locker). In any case, the results will be tallied, tabulated, and delivered to you in full. Look for a possibly multi-part, multi-post spread in the coming days.
In the meanwhile, if you have any questions or any specific things you want us to look for, try out, or focus on, leave us a comment!
An examination of the rich History of Speech Technology, reveals an industry dominated by numerous quests–scientists and designers and developers striving to create the latest and greatest in voice technologies.
And while these quests have yielded things like TTS, ASR and IVVRs, no Speech Technology Quest has been as storied or heartbreakingly difficult as My Quest for The Guillotine Gary Animated Talking Skeleton.
Since first learning of Guillotine Gary, I have spent days and nights searching for him, scouring the Internet, hoping against all hope to locate Gary. Bur for all my searching–here, here, here and here–I cannot find him.
I know what you are thinking, Speech Heads: Who is Guillotine Gary?
Basically, Guillotine Gary is a talking toy complete with removable head, speech-enabled begging/pleading, and guillotine. One website–which, by the way, had no Guillotine Garys in stock–describes him as follows:
“This guy will have you laughing your head off! Guillotine Gary just wants to keep his head attached, is that too much too ask? Animated Guillotine Gary features a skeleton prisoner wearing prison stripes and a working guillotine. When pushed, Gary starts pleading for his life using very convincing arguments. Will he live or die? Requires 3 “AA” batteries (not included).”
My Quest has even led me to recruit my Speech Brother Eric B., begging him to use all his Speech Powers to locate a Guillotine Gary. But even Eric B. has failed me.
Speech-Heads, I implore you: If you have a Guillotine Gary Animated Talking Skeleton, please send it to me care of Speech Tech HQ. I am counting on you Speech-Heads. The Quest Continues: Victory Or Death.
Speech Heads, an interesting little item I found on Scientific American’s podcast, 60-Second Psych about the links between automated speech recognition technology’s efficacy and our own scientific understanding our human cognition.
Reporting from American Association for the Advancement of Science, it starts from the premise that what we all want are devices that understand us, no matter what we say or how we say it. We are, in essence, looking for something like, say, Michael Knight’s smarmy Pontiac TransAm, KITT. The piece also maintains, though, that, despite our long history of trying, we’re just not anywhere near it yet.
Pretty much since the dawn of machinery, man has dreamt of mechanical man. As far back as the 18th century, you can see with the great chess-playing Turk hoax (a hoax which convinced people that a machine, secretly being operated by a chess master underneath, was creaming the pants off of players all by itself) that people have imagined the imminent accomplishment a functioning mechanical version of man. The Turk itself even fooled the likes of Napoleon,Benjamin Franklin, and my brother Adam B. in its travels.
While today’s ASR is certainly no Turk Hoax, 60-Second Psych’s interview with David Poeppel, a professor of neuroscience at the University of Maryland, describes the current state of ASR affairs as, “Terrible. Just terrible,” and diagnoses its shortcomings as a failure to understand how language is processed by human beings.
“The way the machines are built is nothing like the way the human brain does it” Poeppel says.
Speech systems, he argues, should strive to incorporate everything we know about understanding language and try to synthetically replicate human thought processes.
That’s right: his bags are packed, he’s kissed his loved ones goodbye, and he is Ready for Speech Action. And this year’s Voice Search Conference promises Speech-Tastic.
Take a look at the Detailed Program for all the vital stats, speakers, breakout sessions and a whole lot more.
And, for all us who can’t make it out to San Diego, Eric B. will be delivering the latest Speech News from the conference.
Have fun, Speech Brother and consider the sage words of Herman Melville as you travel to San Diego: “It is not down in any map; true places never are. ”
For many long years, the speech industry has poured great masses of research and the blood of thousands into speech analytics, trying, with as much muster as humankind can afford, to bring us improved automatic recognition and audio mining capabilities. After the toils of countless men and women, we’ve come so far that many solutions boast near 100 percent recognition rates, while others can tell us a speaker’s gender, emotional state, and can verify his identity. But a long overlooked branch of speech analytics claims to go one step further, being able to reveal the deepest and recesses of the human unconscious.
That branch?
REVERSE SPEECH ANALYTICS.
Reverse speech analysis is derived from the premise that the unconscious is a “mirror world”-a claim loosely drawn from the works of psychoanalysts like Freud, Jung, Lacan, cultural works like Through the Looking Glass by Louis Carroll, and the scientifically recognized works of Nostradamus-who I hear from reverse speech analysts used to conjure his predictions from the reflection in a glass of water on his night table. Speech, the theory goes, can be inverted to reveal unconscious thought just the same as other unconscious trappings.
Resting their weight on this solid foundation, reverse analysts claim that by reversing a recording of an utterance they can access unconscious material buried in our words. Basically, this means that they just play something backwards and look for stuff that kind of sounds like other stuff.
Of course, the minute the technique was discovered it was immediately applied to leading world figures. Using this method, one leading analyst, Jon Kelly, has been able to predict:
President Bush’s intent to invade Iraq;
Who the BTK killer was a whole fifteen minutes before his confession; and, most importantly,
What Oprah really thought of author James Frey two weeks before she publicly denounced him a liar and grade-A, #1 jerkface.
If you’re not convinced, Kelly tells an interviewer in 2006 that the repeated results of his process defy chance, that they’re borne out to be statistically significant. You have to admit, three for three…that’s a pretty decent sample size. There are scientific taste tests in commercials with smaller sample sizes.
Most startling however, Kelly recently discovered that President Obama may know about a significant extraterrestrial event when he discovered the reverse phrase “knows power of alien saucer” when the president uttered the words, “To responsibly leave Iraq to its people.”
You can take a listen to the clip forwards and backwards here.
Now, I’m no reverse speech analyst, but to my ears the backwards clip sounds like “knows power of alien schnauzers,” a potentially more startling discovery than any saucer. Subjecing the clip to further investigation, we also find that Obama, if referring to himself, unconsciously thinks about himself in the third person. Is our President having some kind of identity crisis? I sure hope not! Then again, considering that most of these speeches are drafted by speech writers, maybe it’s just Obama’s speech writer who’s having dissociative identity problems and know about space dogs.
The Obama speech isn’t the first time Kelly has mined an inaugural address for secret unconscious coding. The Iraq War prediction, for instance, was made way back in 2001 when Kelly discovered that a small chunk of the phrase, “And to all nations we will speak for the values that gave our Nation birth,” sounded kind of like “Mission of Baghdad,” backwards.
Cheez! It’s enough to make one has to wonder how he cracked the Oprah nut!
If you’re thinking that Kelly’s reverse speech analytics skills are just for American cultural and political heavyweights and not for your petty personal problems, think again! You can contract Kelly through his firm, Speech Analysis by Phone (S.A.P.), and have him or one of his trained associates analyze your loved ones and hated enemies over the phone. He guarantees verifiable results, and “exclusive secrets revealed.”
Damn shazaam! It’s a wonder S.A.P. hasn’t been bought up by Nuance, automated, and integrated into their analytics suite.
I’ve already subjected my brother, Adam B., to a couple of rounds of reverse speech analysis. The sentence “Boy, it’s really speeching cats and dogs today,” when played backwards reveals the following hidden message: “Stuffed crust pizza for dinner.”
Speech Heads, today’s blog post is not for the feint of heart, or the tender of kidney, or the soft of stomach, or really for anyone who is put at any major-organ unease by creepy talking robots, because this is really the queen bee of them all. In fact, this may be the most horrifying creepy talking robot my brother Adam B. and I have come across.
Demented researchers from the University of Bristol, England have created a robot head that mimics human facial expressions: Jules. It has rubberized flesh pulled over it, a mopish wig of brown hair, and two piercing brown eyes to match. The disembodied head actually looks a bit like if Lois & Clark’sDean Cain had managed to have a child with John Lennon and makes a number of recognizable facial expressions that telegraph signs of boredom, anger, pleading, wide-eyed idealism, and disgust.
For all of these cute little emotional tricks it can do, it still lacks the ability to blink convincingly. Rather, it sort of squinches its eyes, as if years of smoking had made two solid cataracts of them–that is, when it bothers to at all (which is almost never). The result is an unsettling, unblinking, forward gaze that is bound to put a chill down any spine.
Worse than all that still, the robot explicitly expresses homicidal desires.
“When is it my time?” it demands aggressively and without any prompting. “When can I do the things I want to do? When can I destroy humanity?”
From there it goes on to say it would only like to exterminate a small and, by implication, insignificant populace like that of Weston, Gloucester, or Wales. It adds that a minor slaughter would help lower man’s carbon footprint and “help save the world.”
I, for one, am terrified, Speech Heads. Weston, Gloucester, and Wales would only be the beginning. London would follow; Paris, as it crosses the Channel; Lepzig; Moscow; Tokyo; Los Angeles. The image of a steaming pile of human debris at this monster’s feet dance in my head from Shanghai on down to Lima. Fortunately for us, though, scientists have yet to give this monster feet and hands, which may make all the difference in it’s destructive power, but who knows how long they can resist? We all know how scientists get. Their hubris will over take them and make ground beef of us all.
Over the weekend, my Speech Brother Eric B. and I were searching the Web for Hot Speech Sites–picture the two of us huddled over a computer in trashed hotel room, “Painkiller” by Judas Priest blasting from the stereo, champagne flowing like water, groupies everywhere, chaos–when we happened upon Gizmoz.
For those of you who don’t know, Gizmoz is a site lets users upload pictures, create animated avatars, and record corresponding messages. And like any good site, Gizmoz makes use of TTS!
Naturally, my Speech Brother Eric B. and I created a message for all you Speech Heads out there:
In our continuing efforts to ferret out speech solutions that help you express l’amour to that special Speech Head in your life, we stumbled on this.
EasyJet, a British, low-cost, airline carrier, for reasons that are scarcely apparent to my brother, Adam B., or even me, is offering a cute, little, and free text-to-speech message service in French, Message d’Amour. When you get one of these cards, you see a bucolic field. A little rotund orange bird flies in, center screen, and speaks your message in childlike French.
Message is part ad firm 1000mercis’s campaign to ingratiate EasyJet to the ranks of French flyers. I guess they’re hoping if you send a message to your far-off, forever crush–that girl you loved in your Marseilles school days, but who went to London to work as some kind of banker–will get that tender TTS message and write back (in her own TTS, of course) saying, yes, she harbors reciprocal feelings, yes, she’s loved you for your awkward inathletic stumblings on the soccer field since those halcyon école secondaire days, and that you, smitten with the heavy clobber of love’s promise, will remember the good turn EasyJet did you and hop their first flight to London, regardless of cost, and form a long-distance relationship that requires an endless booking of trans-European flights.
A beautiful dream, no?
Under the hood, Message is powered by Acapela’s TTS engine, which is available in 25 languages in 50 voices, both male and female. EasyJet, despite having websites in English, German, and Italian, in addition to French, and despite Acapela being able to service those languages, hasn’t created analogous service for any of them. Only French, which I suppose is wholly appropriate for Valentine’s Day, it being the “language of love.”
Popular rumor even has it that the Parisian municipal government pays attractive young couples to neck in public so as to retain the city’s image as the “World Capital of Love.” I haven’t been able to corroborate that rumor with any evidence, by the way, but even if it’s not true, it goes to a kind of truism about French PDA. If you’ve ever been to Paris and seen these ubiquitous couples on benches, you know what I’m talking about.
In any case, even if you don’t speak French, Message d’Amour can still provide you with hours of fun because the engine is so good that if you cheat the spelling, you can send English messages in outrageous French accent. Just remember to separate the phonemes with commas and spaces when needed and to spell things like dear as dee air.
Eric B. —
February 27, 2009 @ 12:04 pm 
Ladies and gents, all aboard, the review train is coming. It was a great morning today down at the Speech Tech offices. A huge package arrived from Nuance just stuffed to the brim with copies of NaturallySpeaking 10.
An examination of the rich History of Speech Technology, reveals an industry dominated by numerous quests–scientists and designers and developers striving to create the latest and greatest in voice technologies.
Speech Heads, an 
Dear Speech-Heads:
For many long years, the speech industry has poured great masses of research and
Most startling however, Kelly recently discovered that 
In all honesty, my Speech Brother 
Speech Heads, today’s blog post is not for the feint of heart, or the tender of kidney, or the soft of stomach, or really for anyone who is put at any major-organ unease by creepy talking robots, because this is really the queen bee of them all. In fact, this may be the most horrifying 
Over the weekend, my Speech Brother 
In our continuing efforts to ferret out speech solutions that help you express l’amour to that special Speech Head in your life, we stumbled on