Not sure whether that qualifies as a "real university", but I studied one semester (exchange program) at University of Connecticut. When I arrived there, I remember some people talking to me like I was lucky to get some *real* education for a semester (unlike what I had in my poor country, Canada). Turns out that the level of the classes I got there was so much lower than that of my University (U. of Sherbrooke) that I had to study by myself a before returning for my next semester. It was also the first time I felt I had an impact on the group's average just by myself!
The ones who "ran away with the money" in this case are affectively the ones who sold properties just before the crash. Just like when a bubble or a pyramidal scheme busts, the ones that run with the money are those who got out just before the crash.
This is the main thing I have not yet understood between Canada and Quebec (I'm "Quebecois" but am neither independentist nor federalist): Canada: "you suck, we hate you" Quebec: "We'll leave you" Canada: "Please no, don't leave. We love you" Quebec: "Well, OK then"
They understand the Casimir effect and the related equations, they just can't solve them. So what they do is they find another problem that has the same equations and they measure on that system. If both systems behave using the same equations, then the result should be the same.
Don't know about any refund, but when I was in Kyoto, I remember that when the (Keihan) train wasn't on time, it was usually because I had to adjust my watch (and I would often adjust my watch at the time the train left the station).
I don't know about where you live, but where I live (Montreal, Canada), bus schedules would certainly qualify as creative work in the "fiction" category.
He is technically correct. Even from our reference frame, the supernova was ~7,500 years old. The time of an event doesn't change with it's location, only with the velocity of the observer. For any observer in the galaxy which is stationary wrt the Earth, the time of the supernova is the same. But for an observer on earth moving close to the speed of light, the time of the event is different from ours.
Please, this "how can just 4600 people represent so many" comment is something any college-educated person should know better than to say. Provided the sample was drawn randomly from a representative pool of users, 4600 people is more than adequate, giving a sampling error of about 2%.
The sample size is (barely) enough for a claim that 99% of users have flash. But the main problem is the drawn randomly from a representative pool of users part. There's just no way to sample things like that randomly and that's where the problem is.
No it would not. It would help a bit, but it's not enough. You should have access to the media files you buy, not just the ones for which someone has figured out a way break the DRM.
Algorithmic delay for AAC-LC is just a 20ms, and MP2 is just 35 ms.
Neither AAC-LD (not LC), nor MP2 can give you good speech quality at 16 bbps or below. Also the 20 ms for AAC-LD is not what typical applications do once you add the bit reservoir and all (MP2 will have loo-ahead due to psychoacoustics). On top of that, I don't know about MP2, but the complexity of AAC-LD is a *lot* higher than that of typical speech codecs.
One-way delays below 120ms are said to be excellent by most, assuming decent echo cancellation.
That's 120 ms *total* delay. That includes audio buffering on both capture and playback, network latency (speed of light, plus router delay), jitter buffering, transmission time (the time it takes to transfer the bytes in addition to the ping time), encoding time (computation time required to encode and decode). All that gets added up to the codec's delay and a lot of those delays are actually proportional to the frame size.
But you don't have to believe me. Just write your own MP2-based VoIP client and we'll see how well it works. AAC-LD *is* a contender for higher-end applications (but then you should try CELT, but not in the short-term for the simpler VoIP apps.
You're comparing apples and bananas. Speex is a low-rate telephony/VoIP codec. I've never checked the quality of MP2 and HE-AAC (I'd be surprised if MP2 beats Speex on 15 kbps speech), but no matter how good they are, you can't use them in VoIP because they introduce too much delay (frame size and look-ahead are big). I also suspect that the encoder's complexity is a bit too high for some devices. In the same way, Speex isn't at all competing with Vorbis. They just have different applications. Right now, I'm working on a new codec called CELT that can do speech and music. It does none of them as well as Vorbis or Speex but the algorithmic delay is below 10 ms, which means you can even use it for playing music through a network. It's always a trade-off between delay, quality, bit-rate and complexity.
(Note: I'm the author of Speex) As far as I know, the funding is for specific projects, none of which is related to Speex. That's allright because from the point of view of Mozilla, there aren't really any issues that need to be addressed wrt Speex (or FLAC as far as I know). On the other hand, there's a lot of work to do on Theora, which is what (among other things) will get funded. So don't see that as Xiph projects being ignored. The money just goes where important work most needs to be done. And since it's Mozilla founding this, it's normal that the focus is on the web (as opposed to VoIP, which is the main target for Speex).
I don't think it really *required* a constitutional amendment (a law would have been sufficient), but for some reason someone must have thought it would be harder to undo if part of the constitution (I don't live in the US, so I don't know the details).
If a cheap speaker or cheap headphone's frequency response is bad enough to mess with the model's idea of masking, for example, poor quality reproduction can actually make the 'tricks' of MP3 apparent.
Maybe sort of in theory, but I can't seen that happening in practice. What happens in practice is that bad quality hardware will produce distortion that will end up masking the artefacts of the codec, i.e. many subtle details will be lost in the distortion so you won't know whether the encoder coded them properly.
That's great... until the amount of work you're asked to do increases to the point that the best you can hope is get it done in 100% (instead of 200%) of the time. There's just no standard for the amount of work someone should get done because it varies so much from one person to another (can't ask the same from a junior or senior programmer, even though there are exceptions).
Slashdot Mirror
Not sure whether that qualifies as a "real university", but I studied one semester (exchange program) at University of Connecticut. When I arrived there, I remember some people talking to me like I was lucky to get some *real* education for a semester (unlike what I had in my poor country, Canada). Turns out that the level of the classes I got there was so much lower than that of my University (U. of Sherbrooke) that I had to study by myself a before returning for my next semester. It was also the first time I felt I had an impact on the group's average just by myself!
The ones who "ran away with the money" in this case are affectively the ones who sold properties just before the crash. Just like when a bubble or a pyramidal scheme busts, the ones that run with the money are those who got out just before the crash.
I've been told that in some places in the US, the song gets edited to say "And *one* religion too". Kind of changes the intent a bit, doesn't it.
...as pro-evolution content on the web is succeeding at making the creationists go away.
And yet we in Quebec spell exactly the same way as in France. Can't say the same about UK English vs US English vs Canadian English vs Australian, ...
This is the main thing I have not yet understood between Canada and Quebec (I'm "Quebecois" but am neither independentist nor federalist):
Canada: "you suck, we hate you"
Quebec: "We'll leave you"
Canada: "Please no, don't leave. We love you"
Quebec: "Well, OK then"
They understand the Casimir effect and the related equations, they just can't solve them. So what they do is they find another problem that has the same equations and they measure on that system. If both systems behave using the same equations, then the result should be the same.
Don't know about any refund, but when I was in Kyoto, I remember that when the (Keihan) train wasn't on time, it was usually because I had to adjust my watch (and I would often adjust my watch at the time the train left the station).
I don't know about where you live, but where I live (Montreal, Canada), bus schedules would certainly qualify as creative work in the "fiction" category.
He is technically correct. Even from our reference frame, the supernova was ~7,500 years old. The time of an event doesn't change with it's location, only with the velocity of the observer. For any observer in the galaxy which is stationary wrt the Earth, the time of the supernova is the same. But for an observer on earth moving close to the speed of light, the time of the event is different from ours.
Please, this "how can just 4600 people represent so many" comment is something any college-educated person should know better than to say. Provided the sample was drawn randomly from a representative pool of users, 4600 people is more than adequate, giving a sampling error of about 2%.
The sample size is (barely) enough for a claim that 99% of users have flash. But the main problem is the drawn randomly from a representative pool of users part. There's just no way to sample things like that randomly and that's where the problem is.
No it would not. It would help a bit, but it's not enough. You should have access to the media files you buy, not just the ones for which someone has figured out a way break the DRM.
Honestly, I think I haven't yet seen someone receive comments back where they couldn't take a good guess at who the originator was...
Especially when the reviewer's comment says: "you should cite this great work by professor Foo..."
The funny thing is that doing that might cause global cooling because CO2 is much less effective at trapping heat than methane.
Algorithmic delay for AAC-LC is just a 20ms, and MP2 is just 35 ms.
Neither AAC-LD (not LC), nor MP2 can give you good speech quality at 16 bbps or below. Also the 20 ms for AAC-LD is not what typical applications do once you add the bit reservoir and all (MP2 will have loo-ahead due to psychoacoustics). On top of that, I don't know about MP2, but the complexity of AAC-LD is a *lot* higher than that of typical speech codecs.
One-way delays below 120ms are said to be excellent by most, assuming decent echo cancellation.
That's 120 ms *total* delay. That includes audio buffering on both capture and playback, network latency (speed of light, plus router delay), jitter buffering, transmission time (the time it takes to transfer the bytes in addition to the ping time), encoding time (computation time required to encode and decode). All that gets added up to the codec's delay and a lot of those delays are actually proportional to the frame size.
But you don't have to believe me. Just write your own MP2-based VoIP client and we'll see how well it works. AAC-LD *is* a contender for higher-end applications (but then you should try CELT, but not in the short-term for the simpler VoIP apps.
You're comparing apples and bananas. Speex is a low-rate telephony/VoIP codec. I've never checked the quality of MP2 and HE-AAC (I'd be surprised if MP2 beats Speex on 15 kbps speech), but no matter how good they are, you can't use them in VoIP because they introduce too much delay (frame size and look-ahead are big). I also suspect that the encoder's complexity is a bit too high for some devices. In the same way, Speex isn't at all competing with Vorbis. They just have different applications. Right now, I'm working on a new codec called CELT that can do speech and music. It does none of them as well as Vorbis or Speex but the algorithmic delay is below 10 ms, which means you can even use it for playing music through a network. It's always a trade-off between delay, quality, bit-rate and complexity.
(Note: I'm the author of Speex) As far as I know, the funding is for specific projects, none of which is related to Speex. That's allright because from the point of view of Mozilla, there aren't really any issues that need to be addressed wrt Speex (or FLAC as far as I know). On the other hand, there's a lot of work to do on Theora, which is what (among other things) will get funded. So don't see that as Xiph projects being ignored. The money just goes where important work most needs to be done. And since it's Mozilla founding this, it's normal that the focus is on the web (as opposed to VoIP, which is the main target for Speex).
Warning: California causes cancer.
Five years after being passed, the law banning flies still hasn't reduced the amount of flies. What went wrong?
If we don't extend copyright, what incentive will dead authors have to create?
I don't think it really *required* a constitutional amendment (a law would have been sufficient), but for some reason someone must have thought it would be harder to undo if part of the constitution (I don't live in the US, so I don't know the details).
If a cheap speaker or cheap headphone's frequency response is bad enough to mess with the model's idea of masking, for example, poor quality reproduction can actually make the 'tricks' of MP3 apparent.
Maybe sort of in theory, but I can't seen that happening in practice. What happens in practice is that bad quality hardware will produce distortion that will end up masking the artefacts of the codec, i.e. many subtle details will be lost in the distortion so you won't know whether the encoder coded them properly.
Then take a 64 kbps test and encode your music at that rate. When you want higher quality, you use 128 kbps and you need to test that more carefully.
That's great... until the amount of work you're asked to do increases to the point that the best you can hope is get it done in 100% (instead of 200%) of the time. There's just no standard for the amount of work someone should get done because it varies so much from one person to another (can't ask the same from a junior or senior programmer, even though there are exceptions).
10KW of electricity produce 7KW of heat
And what happens to the other 3 kW? It's lost as heat?