I hate to say it, but maybe the world would be better off if we spent the billions of dollars used on the air attacks (and now "cyberwar") on a crash program to just assasinate Milosevic. Instead of killing innocent civillians, often children, let's take out the one person who made ethnic clensing in Croatia, Bosnia, and now Kosovo possible.
Some people might argue "that wouldn't change anything, worse people might come into power." Then take them out as well. Take full responsibility. Hold leaders personally responsible.
If we're going to fight dictatorial tyrants, we should fight THEM, not the people they send into war.
This obviously leaked story is FUD itself - IS managers see that Microsoft is getting serious about killing Linux. We better not install Linux if Microsoft is about to kill it.
On tuesday, an amendment banning distribution of information on making explosives, destructive devices, and weapons of mass destruction was added to the sure-to-pass Juvenile Crime Bill.
Amendment SP353 to S.254
Sec 402 Criminal Prohibition on Distribution of Certain Information Relating to Explosives, Destructive Devices, and Weapons of Mass Destruction.
(2) Prohibition - It shall be unlawful for any person - (A) to teach or demonstrate the making of an explosive, a destructive device, or a weapon of mass destruction, or to distribute by any means information pertaining to, in whole or in part, the manufacture or use of an explosive, destructive device, or weapon of mass destruction, with the intent that that teaching, demonstration, or information be used for or in furtherance of an activity that consitutes a Federal crime of violence;
[The Senate doesn't have the guts to have this amendment on their site yet, go to the GPO Congressional Record and search on 1999 CR for S5385, then look in that result for Section 402 for the down and dirty]
First of all, MS Audio at 20kbps blows away any other music audio codec I've ever heard at the same bitrate. It sounds like 64kbps MP3. I think Real will need to play catch up here.
Modern music audio codecs are VERY complex. They begin with frequency transformations, then use perceptual modeling to figure out which frequencies of the audio signal are "masked" by other louder frequencies. Then you quantize the frequency coefficients using Huffman or arithmetic coding. The way you quantize is very important and takes a lot of tweaking. Then comes linear prediction. It's tough stuff! Not that it couldn't be done by a CS grad student though.
I'd like to suggest Americans go to Beograd.Com and click on the Java Chat. Go tell them you are from the U.S. It is an interesting experience to "hear from the enemy."
First of all, all this death-of-MP3 stuff is written to shock people. I believe that is why this article was written, and why Mark Cuban of Broadcast.Com has said something similar recently.
However, that said, Twin-VQ (vector quantization) technology, aka VQF may cut audio file sizes in half while maintaining a similar audio quality.
On audio quality, MP3 only approaches CD quality at 128kbps. However lots of MP3 is encoded at lower bitrates, but it is just fine for many applications (such as listening in the office on crappy headphones).
I used to work on CMOS analog neuromorphic VLSI (at Hopkins for a while, no less). The field was very interesting back in the days when microprocessor speeds were at 66 MHz, but in today's 500 MHz plus world, there is very little benefit in neuromorphics except for mobile micro-robots, where power is a concern.
Every other neural AI application is much better served using software running on high-speed digital machines. Of course, there are very, very few useful neural network applications period. We have gallantly figured out how to make neural nets learn, but they are not leading to the kinds of significant AI that we had originally thought.
Mark told the webcasting mailing list that scaling meant:
"how economically and gracefully a server delivering a stream scales...its easy to deliver 100 simultaneous streams, it gets a lot harder delivering 100k simultaneous streams and more. So its how the server scales, how the server interacts withother servers and with users and what kind of programming interfaces are available to enable all of this"
I believe that Mark is thinking in his own terms of Broadcast.Com, which needs to be capable of serving a potential 100k streams. MP3 would have to be packaged in something like Shoutcast or Icecast that is dependable and scalable to this level. It isn't...yet.
Whether Mark is also trying to just poo-poo an explosion in webcasting that didn't come from Broadcast.Com is left up to the reader:) But he does have some serious concerns.
There is an MP3 plug-in for RealPlayer G2, and RealNetworks is talking about making it more tightly integrated. This may be the scalability answer.
Truth be told, I don't think MP3 is the best solution for large, live events. It is a great solution for music-on-demand.
Mind you, MP3 is now an old codec. New vector-quanitzation technology (such as TwinVQ aka VQF) is going to make MP3 quality music available at lower bitrates. The question is whether VQF will be licensed in a way to make it "the next MP3" or "the next forgotten music codec."
MPEG4 multimedia standard will include several audio codecs, including as low as 2kbps speech, VQF, and high-bitrate high-fidelity music codecs as well.
For the last few years, I have must have heard ten or twenty people talk about mining the zone files for marketing information. I can only assume that some of these obnoxious people have gone ahead and done just that.
I'm curious, though, if NSI will now open up a "service" for companies that want the files...
I think the next major move in computer platforms is in wearables/ultraportables, capable of web browsing and email with radio links (such as Ricochet).
Full access from everywhere is the next trend. The limited Palms won't be enough.
I was a programming professionally before I went into college. What I did learn in college is (1) alot of math that I'd never sit down and learn by myself and (2) lots of CS and EE theory, some of which I might have picked up on my own.
However then I went to grad school chasing the elusive spectre of Artificial Intelligence. I remember sitting down with NCSA Mosaic and saying "gee, if it just had cryptographic security, people could do credit card transactions." A few months later, Netscape happened, while I was busy studying for a Ph.D. exam. After that I decided I had to do some part-time Internet work. Eventually I realized that it was time to leave grad school and do Net stuff full-time. Now I'm self-employed and happy!
Slashdot Mirror
Some people might argue "that wouldn't change anything, worse people might come into power." Then take them out as well. Take full responsibility. Hold leaders personally responsible.
If we're going to fight dictatorial tyrants, we should fight THEM, not the people they send into war.
This obviously leaked story is FUD itself - IS managers see that Microsoft is getting serious about killing Linux. We better not install Linux if Microsoft is about to kill it.
On tuesday, an amendment banning distribution of information on making explosives, destructive devices, and weapons of mass destruction was added to the sure-to-pass Juvenile Crime Bill.
Amendment SP353 to S.254
Sec 402 Criminal Prohibition on Distribution of Certain Information Relating to Explosives, Destructive Devices, and Weapons of Mass Destruction.
(2) Prohibition - It shall be unlawful for any person - (A) to teach or demonstrate the making of an explosive, a destructive device, or a weapon of mass destruction, or to distribute by any means information pertaining to, in whole or in part, the manufacture or use of an explosive, destructive device, or weapon of mass destruction, with the intent that that teaching, demonstration, or information be used for or in furtherance of an activity that consitutes a Federal crime of violence;
[The Senate doesn't have the guts to have this amendment on their site yet, go to the GPO Congressional Record and search on 1999 CR for S5385, then look in that result for Section 402 for the down and dirty]
First of all, MS Audio at 20kbps blows away any other music audio codec I've ever heard at the same bitrate. It sounds like 64kbps MP3. I think Real will need to play catch up here.
Modern music audio codecs are VERY complex. They begin with frequency transformations, then use perceptual modeling to figure out which frequencies of the audio signal are "masked" by other louder frequencies. Then you quantize the frequency coefficients using Huffman or arithmetic coding. The way you quantize is very important and takes a lot of tweaking. Then comes linear prediction. It's tough stuff! Not that it couldn't be done by a CS grad student though.
I'd like to suggest Americans go to
Beograd.Com and click on the Java Chat. Go tell them you are from the U.S. It is an interesting experience to "hear from the enemy."
Keep in mind that MP3 is not actually an "open" system as in "open source software." You still have to license it from Fraunhofer/Thomson.
Links to a Mac version of Yamaha SoundVQ can be found at VQF Kingdom.
Currently there is NOT a Linux version, perhaps a Japanese-speaking Linux nerd can talk with the folks at NTT about licensing TwinVQ.
There is a hardware VQF player coming out in Japan in fall 1999 called SolidAudio. You can read a review of the prototype here.
First of all, all this death-of-MP3 stuff is written to shock people. I believe that is why this article was written, and why Mark Cuban of Broadcast.Com has said something similar recently.
However, that said, Twin-VQ (vector quantization) technology, aka VQF may cut audio file sizes in half while maintaining a similar audio quality.
On audio quality, MP3 only approaches CD quality at 128kbps. However lots of MP3 is encoded at lower bitrates, but it is just fine for many applications (such as listening in the office on crappy headphones).
I used to work on CMOS analog neuromorphic VLSI
(at Hopkins for a while, no less). The field was very interesting back in the days when microprocessor speeds were at 66 MHz, but in today's 500 MHz plus world, there is very little benefit in neuromorphics except for mobile micro-robots, where power is a concern.
Every other neural AI application is much better served using software running on high-speed digital machines. Of course, there are very, very few useful neural network applications period. We have gallantly figured out how to make neural nets learn, but they are not leading to the kinds of significant AI that we had originally thought.
Mark told the webcasting mailing list that scaling meant:
:) But he does have some serious concerns.
"how economically and gracefully a server delivering a stream scales...its easy to deliver 100 simultaneous streams, it gets a lot harder delivering 100k simultaneous streams and more. So its how the server scales, how the server interacts withother servers and with users and what kind of programming interfaces are available to enable all of this"
I believe that Mark is thinking in his own terms of Broadcast.Com, which needs to be capable of serving a potential 100k streams. MP3 would have to be packaged in something like Shoutcast or Icecast that is dependable and scalable to this level. It isn't...yet.
Whether Mark is also trying to just poo-poo an explosion in webcasting that didn't come from Broadcast.Com is left up to the reader
There is an MP3 plug-in for RealPlayer G2, and RealNetworks is talking about making it more tightly integrated. This may be the scalability answer.
Truth be told, I don't think MP3 is the best solution for large, live events. It is a great solution for music-on-demand.
Mind you, MP3 is now an old codec. New vector-quanitzation technology (such as TwinVQ aka VQF) is going to make MP3 quality music available at lower bitrates. The question is whether VQF will be licensed in a way to make it "the next MP3" or "the next forgotten music codec."
MPEG4 multimedia standard will include several audio codecs, including as low as 2kbps speech, VQF, and high-bitrate high-fidelity music codecs as well.
BTW, if you want to read more by Declan McCullagh, the guy who broke the story, check out Y2K Culture.Com or see him in video on Meeks Unfiltered
For the last few years, I have must have heard ten or twenty people talk about mining the zone files for marketing information. I can only assume that some of these obnoxious people have gone ahead and done just that.
I'm curious, though, if NSI will now open up a "service" for companies that want the files...
Let's not forget about the EU decision to make web and media caching illegal without prior permission (Wired Article).
I hope that Europe wises up to the need for Internet free speech.
If anyone has video of the Windows Refund Day,
I'd be glad to encode it and put it on the Net
in streaming format - pop email to info@thesync.com
I think the next major move in computer platforms is in wearables/ultraportables, capable of web browsing and email with radio links (such as Ricochet).
Full access from everywhere is the next trend. The limited Palms won't be enough.
I was a programming professionally before I went into college. What I did learn in college is (1) alot of math that I'd never sit down and learn by myself and (2) lots of CS and EE theory, some of which I might have picked up on my own.
However then I went to grad school chasing the elusive spectre of Artificial Intelligence. I remember sitting down with NCSA Mosaic and saying "gee, if it just had cryptographic security, people could do credit card transactions." A few months later, Netscape happened, while I was busy studying for a Ph.D. exam. After that I decided I had to do some part-time Internet work. Eventually I realized that it was time to leave grad school and do Net stuff full-time. Now I'm self-employed and happy!