It's all true, except imo for the bit about the natural evolution of language being a bad thing. I don't mind Japanese use of loan words with katakana except that it often gives Japanese the false impression that English-speakers should be able to understand English pronounced through katakana.
What are the alternatives to movieok, anyway? karamovie? karamuubi? karabi? emptyga? emga (yikes, sounding like enema)?
As for 'Movie Orchestra' -- personally I'd love to get my hands on a copy of Lord of the Rings without Howard Shore's dire score and try my hand at giving a decent soundtrack using existing material or by writing new stuff. I guess that isn't much of a bar game, though...
English loan words in Japan can be funny like that, taking on a meaning all their own. 'My' in Japan now means "one's" or "personal", and is used even when refering to other people, as in:
"Jun, you have a 'my car', right?"
=
"Jun, you have your own car, right?"
But isn't the (scientific) fight against malaria largely won? There are some strains of malaria that are drug resistant, but new drugs are quickly developed. The problem, as far as I understand it, isn't so much that we don't have effective drugs to fight malaria, but that they're simply not affordable to the third world where people need them the most.
It's a nice step, but doesn't help with the main problem - that a significant portion of HIV cases are caused by ignorance and misbelief, especially in Africa.
I would think that the main problem is actually that AIDS is killing millions of people directly and ruining hundreds of millions of lives indirectly as the economy suffers or they lose their family and friends. Improved education and a proper cure would of course be better, but if HIV could be made non-life threatening, I think that would be a major advance.
I'm also curious about the details of the vaccine, though. Is it administered once or is it necessary to administer it at regular intervals? What are its side-effects? I guess it's quite early, but I look forward to hearing more about it.
I don't know the nitty gritty of mp3 implementations, but I don't think there should be anything inherently difficult in coding audio which lots of high frequency information. The point with mp3 is not to throw away high frequencies or low frequencies, as such, but to not code with as many bits those frequencies which are inaudible due to the masking effects of other loud components.
I think the problem that has generally plagued psychoacoustic coders has not been the masking calculations, per se, but rather the frame size they use -- if they transform 1024 samples into the frequency domain but the first 512 samples were really quiet and the next 512 samples represent the transient of a drum or something, then it is common for some pre-echo to occur, post coding, when the signal is transformed back into the time domain. Since classical or jazz recordings would tend to have greater dynamic range and more of these transients, they would generally suffer more due to this process, though coders try to minimize it by dynamically choosing frame sizes according to the content of the frame.
That said, I think that a lot of rock music has crazy digital processing that makes left and right channels fairly uncorrelated, so mp3 codecs which try to do joint-stereo coding or whatever will tend to sound worse (in my experience) on highly processed rock or techno music then they will on less drastically processed acoustic recordings
My feelings are that it's all pretty personal and unpredictable even given the same listening subject -- some noise might distract a listener more than another type of noise, or noise which might be imperceptible at first might become annoying over time, or...
I agree, and if you go here you can find an analysis which considers the molecular size of the PVC polymers in determining the vinyl dynamic range. The result?
(after determining the SNR for the grooves made on a diamond disc to be around 110 dB)
PVC is a Polymer. This means its molecules have been grown by joining together lots of smaller molecules. The results of this polymerization process will depend upon the details of the process. The average molecular weights of the polymer chains which are formed can range from a few tens of hydrogen atom masses to hundreds of thousands. As a result, the PVC molecules are much larger than carbon atoms. This has the effect of producing a material which is 'lumpy' with a typical quantisation size far bigger than a carbon atom. As a result, the value for we should have used for the above expressions is hundreds of times larger than 0.5 nm, producing a much smaller dynamic range. As an example, if we assume the molecules in LP Vinyl are 100 times larger than a carbon atom, then resulting dynamic range might be expected to fall by 40dB to around 70dB.
The purpose of the above example was to help us recognise that, since LPs are made from a collection of real molecules, the signals they hold must be quantised. Fortunately for the LP this usually isn't obvious. The underlying signal quantisation is usually masked by various effects.
A quick look on google turned up this analysis of prestine vinyl dynamic range.
The summary: vinyl can be expected to have a dynamic range of around 80dB to 90dB new, which is clearly worse than CD's 96dB. At frequencies above 1kHz, the dynamic range may be greater than 100dB, so perhaps some songs will play better on vinyl than CD (to begin with, anyway), but as a role, CD is better.
I agree that the quality is well short of FM, but if you compare the analog shortwave samples on that site to the same signals being sent digitally, there's no comparison between the two -- the digital shortwave is much clearer.
It still sounds like less than 64 kbps MP3 quality at its best, though, and more like low bitrate realaudio at its worst (ie, the last sample pair in the list).
Why in the world does anyone think that identity depends upon someone's face? Are people really that simple-minded?
Well, it's a question that alot of people intelligent people have pondered. Notably, think of the late Japanese avant-garde author Kobo Abe and his novel
Face of Another in which the Abe explores the role of masks in determining self and one's interactions with society through the fictional diary of a scientist who loses his face in a horrible laboratory accident and has it replaced with a synthetic mask made based on the specifications of a stranger. The same novel was made into a movie in 1966 by the late and great director Hiroshi Teshigahara.
Then as well of how people often feel uninhibited when they wear masks or paint their faces -- be it at a masquerade or before going to war. Having one's face replace following in accident may not be as deliberate an act, but if the new face offers anonymity and, through people's different responses to one's presence, a different view on the world, is it really so hard to believe that it might to some extent change the identity of the wearer?
Like you say, the DSP effects try to model the real-life non-linearities. A much bigger problem than the cumulative rounding errors, which basically just lower the SNR (not so great on a lot of vintage equipment, anyway) is the internal sampling rate. To mimic vintage equipment requires the implementation of the non-linearities present in those pieces of equipment. Implementing a non-linear stage in the discrete-time domain will, however, always create aliasing distortion -- so the trick is to oversample internally to the point where the folded back aliased frequencies are attenuated enough not to be audible to the human ear.
You can imagine, though, that one can make all sorts of interesting algorithms for generating the full mix from the six string outputs; since they'd be independetly captured digitally, they could then be used to frequency-transform each other or do any number of other bizarre things -- ultimately making sounds that are nothing like a guitar, but still take advantage of the guitar's expressivity. I'm not quite sure what algorithms they could use, but the extra degree of freedom could be quite exciting. And at least the guitar isn't what I thought it was going to be (a physically modelling guitar) -- it still plays like normal, beginning with the vibration of real strings.
You sent shivers down my spine just mentioning Hypnautic Hammer. Despite its outrageously pompous preamble, the extensors actually did change my life with that demo.
However at the same time there are parents with seven-year-old children that are more then able to handle this movie. It's the parents' choice and don't try to blame someone else because of your parenting decisions.
That's all very fine, but when I saw parents taking their 7 year-olds (and younger) to see Saving Private Ryan it became pretty clear to me that parents can not be trusted to act with the best interests of their children and theatres cannot be trusted to trust the parents.
That 2% figure does seem pretty suspicious -- are you sure that it doesn't represent the number of fluent English speakers?
Anyway, I'm sure local language support in India is a good idea, but since the literary community pretty much already uses English as the defacto national standard (or that's what I've read, anyway), what's the big deal with English being the standard for computing? While a majority of the population may not be fluent in English, Indian educators doesn't seem completely incapable of teaching English (unlike their Japanese counterparts) and Indians seem keen to learn it.
If Hindi actually is more practical, sure, why not use it?- but post-partition attempts to establish Hindi as the official national language failed and I don't see why it would suddenly be embraced now.
Not only do they not understand it, but they also mod each other up so that the only immediately visible posts are the ones that are way off base. Bah.
It may not have physically passed on its traits to any offspring, but from the sounds of it the program did internally pass on traits to the next generation (ie iteration of the program) when those traits proved to be successful. That's how an evolutionary/genetic algorithm works, and while it may not be evolution in the biological sense of the word, it clearly models the biological process.
If you want to know WHICH codec will sound the best at 128kbit, you should look at which codec sounds the best at 64kbit--the two are likely to be the same.
Many codecs, while including an ability to degrade 'gracefully', are optimized for a certain bitrate. There isn't just no guarantee that the one which degrades the most gracefully will have the best quality at higher bitrates, but in fact the opposite is generally true; codecs which are designed to degrade gracefully use algorithms that are often inefficiently designed for higher bitrates. I've read through a lot of papers describing codecs that are applied over a wide range of bitrates, and without fail (as far as I can remember) the tradeoff to these codecs was that they had poorer performance at higher bitrates (128+kbps). This inefficiency at high bitrates might be avoided if a multi-mode codec were used that used different algorithms for higher bitrates, but that would again invalidate any extrapolations made from lower bitrates.
I do agree that listening to a particular algorithm at a low bitrate may be a useful way of learning its characteristics, but you cannot reliably infer from these results its performance relative to other codecs at high bitrates.
I agree that each song should have been evaluated by every test subject, but I think it's completely valid to include Green Day or music for which the original recording was of mediocre quality; the purpose of the test was obviously not to see which codec had the best transient response, lowest weighted snr or what-have-you -- the purpose of the test was to determine which codec was subjectively preferred under realistic low bitrate application of the codecs.
People encode Green Day and the Mamas and Papas (frighteningly enough), so those artists are fair game.
Spider-Man and Batman also have a uniquely American and, until September 11, old-fashioned sense of civics... "With great power comes great responsibility"
I disagree with that statement - there's nothing 'uniquely' american about recognizing that great responsibility comes with great powers. But I'm not the only one who disagrees - while Sam Raimi may have just been blabbing on with no point, he said quite seriously in an press conference for Japanese TV (shown on CSN1 in Japan) that while he thought Batman was a very American hero, he felt that Spider-Man belonged to the world.
Of course, he could just be interested in overseas cash for his flick, but he did say it and I think it's worth mentioning.
Aside from the fact that Yukio Mishima was Japanese and the Sea of Fertility did end as the author of the piece describes, what's the connection with Yukio Mishima? Or was that just some academic name dropping?
Any search should turn up lots of information on Yukio Mishima, but a short and rather jumbled summary follows:
Yukio Mishima commit ritual seppuku in 1972 at a Self Defence Forces base after finising the fourth book of the Sea of Fertility tetrology that morning. He seized control of the base with his private ultra-rightest army but, failing to convince his audience at the base of the need to return to the nationalistic heroic ideals of pre-war Japan,he cut open his stomach and an assistant decapitated him.
He was a homosexual, but very much obsessed with esthetics and traditional society. His novels which dealt with his own struggles between heroic idealism, homosexuality and aging disallusion were highly acclaimed in Japan and abroad, though you would be hard pressed to find many contemporary Japanese lay people under the age of 40 who have read his novels.
While many people find the software to be useful, Gator also has built in some more questionable features.
Who exactly finds Gator to be more useful than it is annoying? And does it provide any features (rembering passwords - !!wow!!) which aren't offered by Mozilla?
I accidentally got Gator piggybacked on audiogalaxy a while back, though at the time I didn't clue in that that was where it had come from. I half uninstalled it, but some of it yet remains. Perhaps I'll make a dummy gator.exe file like the other poster suggested.
I'm not entirely clear about what your reservation is. If you're saying that an arrangement, programmed on a computer, can never capture human interactions or unpredictability, then i agree. I would also agree that most programmed music lacks that spontaneity -- but I would even have reservations about that, since I have heard professional arranged midi-based files which (at low bitrates, mind you) rivalled orchestral versions of the same songs. (Never mind that I have also listened to music by the Black Dog which surprised me at every turn)
Most of all, I wonder why anyone would listen to computer arrangements of the sorts of songs which benefit the most from human inflection; I would hardly hold it against computer compositional and synthesis software that it cannot mimic monks chanting. What computers can do, however, is allow musicians to explore all sorts of other avenues which cannot be created by chanting monks or even gifted jazz players, and in the best of all worlds the chanting monks would take advantage, where appropriate, of interactive software (like pure data/max-msp/nato.0+55) to realize whatever sort of sonic texture they wanted while maintaing whatever sort of spontaneity they felt they needed.
Moreover, while computers recreate perfectly whatever you enter, there's still plenty of room for serendipitous results in patches that one can never fully predict our results that one could never have completely foreseen until they are programmed; particularly in the more abstract compositional programs, it is rare that what you programmed is exactly what you had heard in your mind beforehand.
If you're talking about all in one solutions, I would think that highly-programmable software such as pure data, being free and fairly (?) open would top the list, along with less open, but also powerful, packages like MAX/MSP. And if you're talking about Reason, I would think that all-in-one (cheap) packages such asOrion would deserve a mention.
I don't really use (beyo0nd experimentation) any of that software, though - sticking to my own buggy stuff and my hardware synths - so I'm no expert - but next time I update my own (very limited and crash-prone) software synth, it will certainly be a DirectX instrument and maybe a pure data object.
Slashdot Mirror
It's all true, except imo for the bit about the natural evolution of language being a bad thing. I don't mind Japanese use of loan words with katakana except that it often gives Japanese the false impression that English-speakers should be able to understand English pronounced through katakana. What are the alternatives to movieok, anyway? karamovie? karamuubi? karabi? emptyga? emga (yikes, sounding like enema)? As for 'Movie Orchestra' -- personally I'd love to get my hands on a copy of Lord of the Rings without Howard Shore's dire score and try my hand at giving a decent soundtrack using existing material or by writing new stuff. I guess that isn't much of a bar game, though...
English loan words in Japan can be funny like that, taking on a meaning all their own. 'My' in Japan now means "one's" or "personal", and is used even when refering to other people, as in:
"Jun, you have a 'my car', right?"
=
"Jun, you have your own car, right?"
But isn't the (scientific) fight against malaria largely won? There are some strains of malaria that are drug resistant, but new drugs are quickly developed. The problem, as far as I understand it, isn't so much that we don't have effective drugs to fight malaria, but that they're simply not affordable to the third world where people need them the most.
I would think that the main problem is actually that AIDS is killing millions of people directly and ruining hundreds of millions of lives indirectly as the economy suffers or they lose their family and friends. Improved education and a proper cure would of course be better, but if HIV could be made non-life threatening, I think that would be a major advance.
I'm also curious about the details of the vaccine, though. Is it administered once or is it necessary to administer it at regular intervals? What are its side-effects? I guess it's quite early, but I look forward to hearing more about it.
I don't know the nitty gritty of mp3 implementations, but I don't think there should be anything inherently difficult in coding audio which lots of high frequency information. The point with mp3 is not to throw away high frequencies or low frequencies, as such, but to not code with as many bits those frequencies which are inaudible due to the masking effects of other loud components.
I think the problem that has generally plagued psychoacoustic coders has not been the masking calculations, per se, but rather the frame size they use -- if they transform 1024 samples into the frequency domain but the first 512 samples were really quiet and the next 512 samples represent the transient of a drum or something, then it is common for some pre-echo to occur, post coding, when the signal is transformed back into the time domain. Since classical or jazz recordings would tend to have greater dynamic range and more of these transients, they would generally suffer more due to this process, though coders try to minimize it by dynamically choosing frame sizes according to the content of the frame.
That said, I think that a lot of rock music has crazy digital processing that makes left and right channels fairly uncorrelated, so mp3 codecs which try to do joint-stereo coding or whatever will tend to sound worse (in my experience) on highly processed rock or techno music then they will on less drastically processed acoustic recordings
My feelings are that it's all pretty personal and unpredictable even given the same listening subject -- some noise might distract a listener more than another type of noise, or noise which might be imperceptible at first might become annoying over time, or...
(after determining the SNR for the grooves made on a diamond disc to be around 110 dB)
A quick look on google turned up this analysis of prestine vinyl dynamic range.
The summary: vinyl can be expected to have a dynamic range of around 80dB to 90dB new, which is clearly worse than CD's 96dB. At frequencies above 1kHz, the dynamic range may be greater than 100dB, so perhaps some songs will play better on vinyl than CD (to begin with, anyway), but as a role, CD is better.
I agree that the quality is well short of FM, but if you compare the analog shortwave samples on that site to the same signals being sent digitally, there's no comparison between the two -- the digital shortwave is much clearer.
It still sounds like less than 64 kbps MP3 quality at its best, though, and more like low bitrate realaudio at its worst (ie, the last sample pair in the list).
Why in the world does anyone think that identity depends upon someone's face? Are people really that simple-minded?
Well, it's a question that alot of people intelligent people have pondered. Notably, think of the late Japanese avant-garde author Kobo Abe and his novel Face of Another in which the Abe explores the role of masks in determining self and one's interactions with society through the fictional diary of a scientist who loses his face in a horrible laboratory accident and has it replaced with a synthetic mask made based on the specifications of a stranger. The same novel was made into a movie in 1966 by the late and great director Hiroshi Teshigahara.
Then as well of how people often feel uninhibited when they wear masks or paint their faces -- be it at a masquerade or before going to war. Having one's face replace following in accident may not be as deliberate an act, but if the new face offers anonymity and, through people's different responses to one's presence, a different view on the world, is it really so hard to believe that it might to some extent change the identity of the wearer?
Like you say, the DSP effects try to model the real-life non-linearities. A much bigger problem than the cumulative rounding errors, which basically just lower the SNR (not so great on a lot of vintage equipment, anyway) is the internal sampling rate. To mimic vintage equipment requires the implementation of the non-linearities present in those pieces of equipment. Implementing a non-linear stage in the discrete-time domain will, however, always create aliasing distortion -- so the trick is to oversample internally to the point where the folded back aliased frequencies are attenuated enough not to be audible to the human ear.
You can imagine, though, that one can make all sorts of interesting algorithms for generating the full mix from the six string outputs; since they'd be independetly captured digitally, they could then be used to frequency-transform each other or do any number of other bizarre things -- ultimately making sounds that are nothing like a guitar, but still take advantage of the guitar's expressivity. I'm not quite sure what algorithms they could use, but the extra degree of freedom could be quite exciting. And at least the guitar isn't what I thought it was going to be (a physically modelling guitar) -- it still plays like normal, beginning with the vibration of real strings.
You sent shivers down my spine just mentioning Hypnautic Hammer. Despite its outrageously pompous preamble, the extensors actually did change my life with that demo.
However at the same time there are parents with seven-year-old children that are more then able to handle this movie. It's the parents' choice and don't try to blame someone else because of your parenting decisions.
That's all very fine, but when I saw parents taking their 7 year-olds (and younger) to see Saving Private Ryan it became pretty clear to me that parents can not be trusted to act with the best interests of their children and theatres cannot be trusted to trust the parents.
That 2% figure does seem pretty suspicious -- are you sure that it doesn't represent the number of fluent English speakers?
Anyway, I'm sure local language support in India is a good idea, but since the literary community pretty much already uses English as the defacto national standard (or that's what I've read, anyway), what's the big deal with English being the standard for computing? While a majority of the population may not be fluent in English, Indian educators doesn't seem completely incapable of teaching English (unlike their Japanese counterparts) and Indians seem keen to learn it.
If Hindi actually is more practical, sure, why not use it?- but post-partition attempts to establish Hindi as the official national language failed and I don't see why it would suddenly be embraced now.
Not only do they not understand it, but they also mod each other up so that the only immediately visible posts are the ones that are way off base. Bah.
It may not have physically passed on its traits to any offspring, but from the sounds of it the program did internally pass on traits to the next generation (ie iteration of the program) when those traits proved to be successful. That's how an evolutionary/genetic algorithm works, and while it may not be evolution in the biological sense of the word, it clearly models the biological process.
You state:
If you want to know WHICH codec will sound the best at 128kbit, you should look at which codec sounds the best at 64kbit--the two are likely to be the same.
Many codecs, while including an ability to degrade 'gracefully', are optimized for a certain bitrate. There isn't just no guarantee that the one which degrades the most gracefully will have the best quality at higher bitrates, but in fact the opposite is generally true; codecs which are designed to degrade gracefully use algorithms that are often inefficiently designed for higher bitrates. I've read through a lot of papers describing codecs that are applied over a wide range of bitrates, and without fail (as far as I can remember) the tradeoff to these codecs was that they had poorer performance at higher bitrates (128+kbps). This inefficiency at high bitrates might be avoided if a multi-mode codec were used that used different algorithms for higher bitrates, but that would again invalidate any extrapolations made from lower bitrates.
I do agree that listening to a particular algorithm at a low bitrate may be a useful way of learning its characteristics, but you cannot reliably infer from these results its performance relative to other codecs at high bitrates.
Who cares what music was used?
I agree that each song should have been evaluated by every test subject, but I think it's completely valid to include Green Day or music for which the original recording was of mediocre quality; the purpose of the test was obviously not to see which codec had the best transient response, lowest weighted snr or what-have-you -- the purpose of the test was to determine which codec was subjectively preferred under realistic low bitrate application of the codecs.
People encode Green Day and the Mamas and Papas (frighteningly enough), so those artists are fair game.
i got a free parking space in fukuoka for anyone who's interested...
Spider-Man and Batman also have a uniquely American and, until September 11, old-fashioned sense of civics... "With great power comes great responsibility"
I disagree with that statement - there's nothing 'uniquely' american about recognizing that great responsibility comes with great powers. But I'm not the only one who disagrees - while Sam Raimi may have just been blabbing on with no point, he said quite seriously in an press conference for Japanese TV (shown on CSN1 in Japan) that while he thought Batman was a very American hero, he felt that Spider-Man belonged to the world.
Of course, he could just be interested in overseas cash for his flick, but he did say it and I think it's worth mentioning.
no kidding - my sharp vcr has the same kind of circular control, too.
Aside from the fact that Yukio Mishima was Japanese and the Sea of Fertility did end as the author of the piece describes, what's the connection with Yukio Mishima? Or was that just some academic name dropping?
Any search should turn up lots of information on Yukio Mishima, but a short and rather jumbled summary follows:
Yukio Mishima commit ritual seppuku in 1972 at a Self Defence Forces base after finising the fourth book of the Sea of Fertility tetrology that morning. He seized control of the base with his private ultra-rightest army but, failing to convince his audience at the base of the need to return to the nationalistic heroic ideals of pre-war Japan,he cut open his stomach and an assistant decapitated him.
He was a homosexual, but very much obsessed with esthetics and traditional society. His novels which dealt with his own struggles between heroic idealism, homosexuality and aging disallusion were highly acclaimed in Japan and abroad, though you would be hard pressed to find many contemporary Japanese lay people under the age of 40 who have read his novels.
Evangelion?
While many people find the software to be useful, Gator also has built in some more questionable features.
Who exactly finds Gator to be more useful than it is annoying? And does it provide any features (rembering passwords - !!wow!!) which aren't offered by Mozilla?
I accidentally got Gator piggybacked on audiogalaxy a while back, though at the time I didn't clue in that that was where it had come from. I half uninstalled it, but some of it yet remains. Perhaps I'll make a dummy gator.exe file like the other poster suggested.
I'm not entirely clear about what your reservation is. If you're saying that an arrangement, programmed on a computer, can never capture human interactions or unpredictability, then i agree. I would also agree that most programmed music lacks that spontaneity -- but I would even have reservations about that, since I have heard professional arranged midi-based files which (at low bitrates, mind you) rivalled orchestral versions of the same songs. (Never mind that I have also listened to music by the Black Dog which surprised me at every turn)
Most of all, I wonder why anyone would listen to computer arrangements of the sorts of songs which benefit the most from human inflection; I would hardly hold it against computer compositional and synthesis software that it cannot mimic monks chanting. What computers can do, however, is allow musicians to explore all sorts of other avenues which cannot be created by chanting monks or even gifted jazz players, and in the best of all worlds the chanting monks would take advantage, where appropriate, of interactive software (like pure data/max-msp/nato.0+55) to realize whatever sort of sonic texture they wanted while maintaing whatever sort of spontaneity they felt they needed.
Moreover, while computers recreate perfectly whatever you enter, there's still plenty of room for serendipitous results in patches that one can never fully predict our results that one could never have completely foreseen until they are programmed; particularly in the more abstract compositional programs, it is rare that what you programmed is exactly what you had heard in your mind beforehand.
If you're talking about all in one solutions, I would think that highly-programmable software such as pure data, being free and fairly (?) open would top the list, along with less open, but also powerful, packages like MAX/MSP. And if you're talking about Reason, I would think that all-in-one (cheap) packages such asOrion would deserve a mention.
I don't really use (beyo0nd experimentation) any of that software, though - sticking to my own buggy stuff and my hardware synths - so I'm no expert - but next time I update my own (very limited and crash-prone) software synth, it will certainly be a DirectX instrument and maybe a pure data object.