-2010 | Recent and Future Ideas | futurist theories -
Casual writings on music, language, science, technology, and human race, in the form of roughly one-paragraph synthesis.
Freakonomics-Cheating, Incentive-based world, correlations and more
We have to applaud for those in the academia who always think out of the box. Steve Levitt is one of such examples. He uses his economics eye to look at various things in the world, and come up with fascinating studies such as "what do high school teachers in the Chicago area have in common with the world of Sumo wrestlers in Japan?", or "what does the decreasing crime rate in the US in the 1990s have to do with the dictator's policy in Romania two decades ago?" And the answer may astonish you. One best thing that I take away from this is that, how we can never be too careful about drawing correlations and causation chains based on simply what we observe from the surface of things--a point that I stressed in a previous blog entry about the water pipe and the switch in my apartment. Finally, also speak of two other things I read and watched lately, Donald Trump's 2008 book, Think Big, Kick Ass, and the documentary movie Inside Job (how Wall Street companies and investment banks have led to the economic crisis one step after another), I have to say that, the world of business (as many other fields) is indeed very complicated and sometimes people get stuck in an awkward position, where they have to do what they have done, although they might know this was not correct. No matter cheating or incentive-based business models, people are trapped in a system that they themselves built and benefit from (and also knowing the problems of the system). It's a communal thing, that everybody knows exactly what's going on, and they had to pretend that everything is going to be alright, because once they challenge the system, or quit, they themselves are the only one who is going to suffer, while others gain on your pain. But it turned out that eventually things are not alright, just like the financial crisis. Thus, to sum up, human beings are selfish animals, just like any other animals without superior intelligence. The only thing they care about is their own survival, not how the world is going to be 20 years later. Donald Trump is right: "Lions in the jungle kill only for food, but humans kill for the fun".
sound based search engine
I have long had the thought of a search engine based on not text, but sound. Imagine you heard a song, didn't know any information about the singer, or lyrics, or album, or record label, and just decided to look it up on the internet. How can you search if you don't have any key words? Well, a sound-based engine will just get you where you want to go. You can search by singing a fraction of what you remembered from the melody, and the likely results of the songs will pop up. The same with speech. Luckily, now there are actually people who are working on this (not singing but speech). Listen to the BBC podcast introducing the speech search technology by a groups of linguists at Oxford University in UK.
I wasn't completely aware how crazy I am about artificial intelligence-robots in this case, until I saw LEGO robot mindstorms in the store. It costs less than $300 and it is programmable from low to high levels. You can also record your own voice and your own image to display on its screen. Wonderful.
Not Exactly....in praise of vagueness
Recently heard this presentation at a banquet at SELLC by Dr. Kees Van Deemter. Rather interesting presentation for attracting our attention from the food at table in Guangzhou. Nonetheless, it is indeed very interesting:
"How warm is a ‘warm day’? Where should we draw the ‘poverty line’? Are you the same person as you were yesterday? None of these questions can be given a clear-cut answer. We operate in a world full of continuous variation, relying on concepts that are not precisely defined but vague around the edges. This book, which cuts across logic, linguistics, and artificial intelligence, considers the challenges posed by vagueness, showing how vagueness is often difficult to avoid, and frequently (though not always!) useful too. It defends a perspective on vagueness that hinges on probabilities instead of crisp dichotomies or degrees of truth. Last but not least, the book shows how Natural Language Generation programs are beginning to come to terms with vagueness, allowing computers to use vague language judicially in their interaction with human users."
music & language--not quite there yet
SELLC is indeed a good place. People here are all about logic and computation. But I also had a chance to familiarize myself with the fact that just how many people have absolutely no idea about the connections between music and language at all, even though they study the logic of language. A lot of questions can't be answered in a simple sentence, and I often find myself being look at strangely and being asked, "what exactly is it that you research about?" and I find myself in an awkward position to give a simple and straight forward answer with a few words.Something interesting happened too, when this American logician Mitch and I talked about singing and speaking: (1) people who speaks with an obvious foreign accent does not necessarily do so in singing (or rather many good singers make you think they're native speakers while they are not; (2) people who stutter in speaking does not in singing; (3) music therapy recovering language function in brain damage patients; (4) speech to song illusion.
music, mind, and machine
There is a MIT Media Lab project with the same title. Now, from the Stanford Laptop Orchestra to the MIT Future Opera, from the Toyota violin-playing robot to the high-quality-classical-music-generation software at UCSB, it seems that music and technology are indeed being connected closely. It seems to me that there is speech synthesis and speech recognition, and there is also singing synthesis (KTH), but music recognition by machine should be available in the future. imagine a machine (or robot) that can identify the music you play just by listening. It will also create a shift in the paradigms of music search engine. Ideally, he can also generate some music to play along with human musicians. Better yet, a improvising robot. Not that different than the classical music writing machine at UCSB, isn't it? Of course, music recognition can be as challenging as speech and face recognition, and the robustness of human perception system is just not as perfectly achieved by machine.
Linguists say language is a skill, rather than a knowledge. Just like a good musician does not need to know music theory (or even musical notation for that matter), a native speaker of certain language doesn't need to be aware of the structure of its language, although s/he has grown to be an expert of using it correctly since childhood. Recently I realized just how unaware one can get. As I was sitting in the classroom listening to lectures about how to teach foreigners Chinese, the first thing is the four tones, and how to pronounce them, etc. Then I came to Guangzhou, a land where Cantonese is the native tongue of millions. I hear Cantonese all the time; I couldn't figure out its tone system, but I knew that it's something like 6+3 tones. So finally I asked a local (who is a grad student in logical semantics), "tell me about the tone systems of Cantonese." "Sorry", he answered, "I have no idea how many tones there are in Cantonese. Absolutely no clue." "How often do you speak?" "I speak everyday since I started talking as a kid."
Any thoughts?Leave a comment without having to log in.
same soul,many bodies
I read this book very accidentally in Singapore in 2008 while I was visiting my dad. He was interested in supernatural things and had a lot of books in this respect. It was in Chinese and published in Taiwan. Just couple of days ago, a colleague at CASS mentioned that he was reading this book recently and had me thinking again. A serious doctor, graduate of Yale and Columbia, M.D, trained in hard science, Dr. Brian L. Weiss was propelled to believe in the theory of same soul, many bodies by the evidence from his psychotherapy. If we do have a soul that recycle itself into many lives, the hypothesis is that the memory of the past life is very very weak and vague and hazy, and deep deep down in your brain. However, I also want to relate this to another mysterious thing, our dreams. I am sitting in my office, and I was typing white letters on the screen of black backdrop. "wow, this is too bright", I thought to myself, and I was trying to change the white text into gray. My colleague just got up and left, and I was looking at the screen--gray, white, black, and orange--and boom!!! Suddenly I felt a strong consciousness that this moment has happened in the past. Striking. This particular moment. With the text, color, colleague, office--this exact moment. Was it in my dream? Usually I should think so. Or maybe it was in my past life. But it wouldn't make sense because I'm wide conscious and there should be no computers in my past life. Logically speaking, there can be a world where there are a more or less finite number of souls (at least much less than our population) with our increasing population today-it almost seems that the addition of a human being is more initiated by the human themselves than a deity or some supernatural force. But I do subscribe to the thought that the ability of our human being to understand this world is very limited. Just I can't understand does not mean it doesn't exist. Here, read it on New York TImes, which suggested that now 25% of Americans believe in reincarnation. (side note: this is also related to children's ability to learn any language in the world. Inborn. UG?)
tech innovation is no easy task
Recently I was reading about Steve Jobs and how he and other guys started the Apple computers and how that turned into what we have today-a digital era where everyone can have a computer on his/her desk and does some form of computing. Although it seems plain today, I was a bit shocked, honestly, to learn that up until the 1980s, even Bill Gates and other Microsoft (and other companies) thought that the computers are just for specialized business, and it's not for personal use. Only when Steve jobs' vision of the Mac computer became available that people began to really see how it can have a market as a layperson thing. I grew up in the 1980s so forgive me for not knowing about the history, because computers seemed always to be part of what's there in the world for us--and even more so for those born in the 1990s. In any case, even today, I cannot imagine that back then how can some one, or some company, have the vision and confidence to be sure about the market of personal computers. Or any innovation for that matter. No matter back then, or today, again, it's still no easy task. Yeah. No one can say for sure. Maybe that's why things can become exciting.
Don't know since when these smart games are beginning to spread their popularity. This one is just like the Plant vs. Zombies, giving you a taste of elegantly built and smartly designed gaming experience that is utterly different from those like Street Fighter or World of Crafts or any of my previous gaming experience (thank you to Zuo Xiaoxuan). In this game you get to learn just how challenging the work is of an architect--building something that is structurally sturdy and also being creative at the same time is not easy. Especially if there are great distances horizontally (like bridges) or vertically (like a high tower). Of course in real world those materials is not the same with the little units in the World of Goo. The soundtrack in this game is also great. Listen or download the music by Kyle Gabler.
tomorrow of speech synthesis
Here is what the speech scientists are working on:(University of York,UK)
Physical modelling voice source for voice synthesis Prof David Howard, email: email@example.com
Today’s electronic voice synthesis is highly intelligible but rarely if ever mistaken for the voice of a human speaker (it is highly non-natural). The next key step in improving the naturalness of electronic voice synthesis (speech and singing) involves creating a virtual model of the oral (mouth) and nasal (nose) tracts that can be varied in shape dynamically according to human articulation. Steps are in progress to achieve this, but in order for it to function in a human-like manner, it requires a voice source that is as human-like as possible, and this is the focus of this PhD. The vocal folds of the human larynx will be modelled in such a way that they will vibrate when a virtual lung airstream is applied, according to the settings of appropriately placed virtual muscles to control pitch and voice quality. The output from the model will be compared to signals that monitor the life outputs from human informants, with the purpose of creating as natural a voice source as possible. This source will be linked to the physical model to establish how natural the resulting voice output is.
human and machine...
"Logic is the only thing that both human and computers understand", remarked Dr. Van Deemter at the banquet of SELLC. It appears that one fundamental question means a lot to some logicians and philosophers, which is, ultimately, can human mind be broken completely into 1s and 0s and we can, like Sheldon Cooper of Big Bang Theory would one day hope, live forever as a machine? One thing was particularly interesting to me: When I talked with H.Peng, a Ph.D of Computer Science at Apple Inc., he talked about his past project of "face recognition", and how good and effortless human is at it and how hard it is for computers to do. I noticed recently that Google Picasa has the face recognition function in its latest versions, where the software can pull out hundreds of photos of your face among thousands of pictures stored on your hard drive. Its performance is indeed quite impressive, although there are occasions that I was mistaken for my high school female classmates. But more strikingly, it does not recognize me in the photos from my childhood; it did a bad job when I make a funny face, or angry face, with my facial structure altered significantly. Meanwhile, we all know how easy it is for human to do it, no matter you're young, sick, wearing different facial accessories, or only showing the left part of your face...So would computers be able to do it like that? I think at least we're all trying to make it do that. The brain is a funny thing. It is an incredible machine, and it is constantly being modified by the very things it created (like music or language). No matter what the implications AI might have for the ego of human intelligence, it is certainly exciting for us to see a personal robot speaking to you, recognizing you, and playing violin with you (like Honda ASIMO).
personal robot & companionship
I recently attended a funeral, which makes me thinking, like Sheldon said, "personal robot cannot get here sooner". One of the top profound sadness of human kind is, for me, the loss of the spouse by the elderly. They are fragile mentally and physically, and just lost their love who they spent most of their lives with. So what if a robot is pre-made, who can speak in the same voice and fashion as the deceased one, and act like him/her in terms of personality? Kind of like a simpler version of "downloading your thoughts from your brain into a machine and live forever as a computer". This could seriously help people get over their difficult times, and it could be a rented option if purchasing is too expensive. After all they may move on one day and no longer need a robot company when a human company can replace it.