I'm not sure what 1995 you're thinking about, but I was buying books from Amazon.com in 1995. I have a single account with a continuous order history all the way back.
There's still a lot of lingering hostility left over from that aquaria debate. For those that weren't tuned in to Usenet the first time around, check out the thread starting here. It's astonishing how worked up gardeners get.
Eric Scheirer, a music technology researcher at MIT's Media Lab, said Freenet is an interesting experiment, but said it would likely be used only by a small community of pirates and "privacy nuts."
I stand by my quote in that article, although naturally it's a little short on context. Let me make clear that I am in favor of privacy, security, and anonymity when appropriate, and I despise the current attempts to make the WWW more corporate-controlled via both code and law. I don't think there's anything wrong with Freenet, I just don't think it will ever take off in the mainstream.
The fact is that most of the things that most people like to use the WWW for--such as e-commerce and Slashdot--cannot be built on Freenet, since it has no cookies and no memory. Given this, I can't see anything happening with Freenet except that it becomes a huge storehouse for illegal porn, pirated MP3s and 3l33t w8r3z.
It's a shame, because the potential political benefits that it raises, by allowing dissident speech in repressive countries, is great.
I guess my question for the Freenet developers would be: I am not a pirate, a privacy nut, a political dissident, or someone trying to spread illegal trade secrets. What does Freenet offer me? And are these benefits broad enough to a broad enough segment of the world population to create the momentum needed for Freenet to work sociologically as well as technically?
I think it is quite clear that all of the problems described here will come to pass, unless the SDMI process changes somehow. The problem as I see it is that unlike MPEG, the IETF, and W3C, which are standards groups attended by geeks, SDMI is mostly attended by suits (lawyers and marketroids). The suits set the agenda and then try to develop a technical standard that fits, even with no real understanding of technology.
I wrote a long column for MP3.com about the dysfunctions of SDMI; you can read it here. The head of SDMI, Leonardo Chiariglione (who is also the head of MPEG), responded in his inimitable manner here.
Two comments (I am not a lawyer and always appreciate being corrected if I get something wrong):
1. It's important to understand the difference between the California case and the 2600 case. The California case is a theft-of-trade-secrets suit, which is unlikely IMHO to succeed. The 2600 case is a suit under the Digital Millenium Copyright Act, which specifically prohibits decryption devices, even when they are constructed through proper reverse engineering. It is much less clear IMHO that DeCSS is not in violation under the DMCA--argument for the defense hinges on the question of whether DeCSS is or is not *necessary*, and has as its *sole purpose*, to acheive interoperability with other DVD players.
2. In some ways lower-court rulings in favor of the DMCA are likely to be of benefit in the long run, because they will accelerate the process of getting higher-court review of the constitutionality of the DMCA. Many legal experts believe that the DMCA is not constitutional. It takes court cases lost in the local jurisdictions in order for this to come out in the legal system. Thus, the most important thing now is to support 2600 and the EFF to continue the fight so that eventually the whole DMCA can be thrown out. Somebody has to be the test case, and it's better if it happens sooner rather than later IMHO. If we won every local case brought under the DMCA due to technicalities, the DMCA and its horrible ramifications would remain in force. Better to lose some small and meaningless fights in order to defeat the DMCA.
This particular fight is about as meaningless as one could be, since there is no practical effect on the Linux or DVD world at large from the ruling. Only the defendants and their contacts are enjoined, so DeCSS distribution is not limited in any important way. Plus I'm sure 2600 is happy for the press coverage.
The MPEG-4 Audio standard is done. It was finished at the October 1998 MPEG meeting and is now "out for ballot", which means the various countries that are members of ISO vote to approve it (countries have had several chances to suggest changes, so it's unlikely to be disapproved). After formal ballot approval, it goes to ISO for publication, and then you can buy it. It will likely be available in final form before the end of the year.
The "reference software" (slow, user-hostile code to demonstrate how the standard is supposed to work) was completed in August and the source is available as part of the spec. Non-MPEG organizations are already building tools for user-friendly use of the standard.
The whole MPEG-4 Audio standard (not including Video or Systems) is about 1200 pages long. It is formally ISO 14496-3:1999 and is divided into 6 Sections:
1. Introduction and Overview
2. Parametric Speech coding
3. CELP Speech coding
4. General Audio (AAC/TwinVQ merger)
5. Structured Audio (audio synthesis)
6. Text to Speech Interface
The part that is most like MP3 is Section 4. Section 4 enables music and other wideband audio coding from 16 kbps to 64 kbps/channel. At the high end, the quality is nearly transparent -- most listeners will not be able to tell the difference between the coded and the original signal. MPEG-4 GA at 96 kbps (stereo) gives about the same quality as MP3 at 192 kbps (stereo) -- thus, files are half as big for the same quality.
There are no "layers" in MPEG-4 Audio.
Some of the sections of the standard (2, 3, 4) are protected by patents and cannot be freely implemented. Section 5 is not protected by patents and can be freely implemented without paying license fees.
Here is the hype from the beginning of Section 1:
ISO/IEC 14496-3 (MPEG-4 Audio) is a new kind of audio standard that integrates many different types of audio coding: natural sound with synthetic sound, low bitrate delivery with high-quality delivery, speech with music, complex soundtracks with simple ones, and traditional content with interactive and virtual-reality content. By standardizing individually sophisticated coding tools as well as a novel, flexible framework for audio synchronization, mixing, and downloaded post-production, the developers of the MPEG-4 Audio standard have created new technology for a new, interactive world of digital audio.
MPEG-4, unlike previous audio standards created by ISO/IEC and other groups, does not target a single application such as real-time telephony or high-quality audio compression. Rather, MPEG-4 Audio is a standard that applies to every application requiring the use of advanced sound compression, synthesis, manipulation, or playback. The subparts that follow specify the state-of-the-art coding tools in several domains; however, MPEG-4 Audio is more than just the sum of its parts. As the tools described here are integrated with the rest of the MPEG-4 standard, exciting new possibilities for object- based audio coding, interactive presentation, dynamic soundtracks, and other sorts of new media, are enabled.
Best regards,
-- Eric Scheirer Editor, ISO 14496-3 (MPEG-4 Audio)
For MP3, Thomson and FhG jointly handle the portfolio of licenses and provide "one-stop shopping" to acquire MP3 licenses (see here ). I don't know exactly how many patents pertain; I would guess more than five but less than fifteen.
For MPEG-4, the licensing authority details are not yet set, but there is general intent to provide a similar one-stop mechanism. I would expect the legal machinery to be in place within a year. There are a lot more patents that apply to MPEG-4 than to MP3, maybe as many as 25 or more (but see below).
Purely in my own opinion, I think that ISO is not nearly strict enough about managing patent claims on the MPEG standards. The basic argument, which is a good one, is that all of the relevant technology is held by patents, and good standard codecs could not be made if patented technology was not admitted. But IMHO the patent-holders are today allowed to abuse their privileges under this system.
Of course there is a point to standards that are protected by patents -- many companies make a lot of money from them in spite of the licenses they have to pay (MPEG-2 video decoders, for example). But these patents also have the effect of making it difficult for hobbyists to develop conforming freeware and open-source products. This may, in fact, be the goal, but it is wrong IMHO.
Not all of MPEG-4 audio technology is protected by patent. The audio synthesis codec that I developed at MIT and contributed to MPEG has been released into the public domain at my request.
-- Eric Scheirer Editor, ISO 14496-3 (MPEG-4 Audio standard)
There's a difference between allowing a single company to control the standard and allowing patents into the standard. Contrary to popular opinion, Fraunhofer is not the only patent-holder on MP3 technology -- they participated in the MPEG process with a number of other companies. They have just been the most aggressive in pursuing infringers. And FhG doesn't control the *direction* of the standard -- only MPEG does that, and it's according to MPEG's normal working rules.
The situation was much different with Java. Sun wanted simultaneously to have Java be an International Standard, but also reserve the right to make unilateral changes to the technology if they decided that it was in their interest to do so.
The difference is clear for MP3 -- no individual company controls the MP3 standard; if revisions need to be made, this is decided by MPEG as a whole, not by Fraunhofer or any other individual company. Sun could not accept letting the other ISO members share in the role of deciding the progression of Java. This is their right, as they own the Java technology, but then the technology can't be an International Standard.
-- Eric Scheirer Editor, ISO 14496-3 (MPEG-4 Audio)
The part of MPEG-4 that's based on QT is the file framework, not the codecs. The codecs in MPEG-4 have no particular relationship to those in QuickTime.
Eric Scheirer Editor, ISO 14496-3 (MPEG-4 Audio standard)
IMHO this is an excellent piece of writing and a valuable contribution. And I think the agenda highlighted for H2O is dead-on.
There's a real question today about the breadth of solutions that can be developed in the open-source model as it currently exists. Can "niche" products attract the developer base that they need to leverage the benefits of the bazaar? This is still unproven, I think.
The folks at Harvard that are involved with this are serious heavyweights when it comes to public policy analysis and lobbying strategy. It behooves the community (IMHO) to think about the issues here sooner than later. Some of us aren't fans of our local or national governments, but it surely must be better for the Government to think that open-source is a Good Thing than that it's a Bad Thing.
The promise of convincing governments of this (obviously, USA-centric at first) is the goal of the organization, and I think it's a worth one.
Look at http://sound.media.mit.edu/~eds/beat/ -- especially under "Audio Morphing". I put this together as a technology demo, but the code is available (contact me) if someone wants to make a GPL'd product out if it.
I wrote an overview of the MP4 audio standard. You can find it here.
Technical work on MP4 was completed in October 1998, and it is in the publication process at ISO now. I'm not aware of any fully-functional encoders/decoders yet.
-- Eric Scheirer Editor, ISO 14496-3 (MPEG-4 Audio)
I wrote a summary of MP4 audio and its relationship to AAC and VQF. Go here.
In summary, some of the technology from VQF is included in the MPEG-4 standard, but a.VQF file is not really an MP4 file since MP4 has its own file format, and VQF is only used in certain restrictive ways within MPEG-4.
The story doesn't mention that Platinum is doing this as part of their process of getting rid of all of their VRML assets and staff. They fired a thousand people on Tuesday, including everyone there working on VRML. *Then* they decided to talk about open-sourcing the Cosmo code.
A majority of the people who are/were heading the projet to create a next-generation VRML standard worked for Platinum and are now without employ. It remains to be seen whether this is the end of VRML development, or whether the VRML community can reorganize itself around open-source development.
A couple of points: There was an independant test comparing PAC, AAC, AC-3, MP2, and some other coders conducted in late 1997 at the Communications Research Centre in Ontario. The results are reported in the March 1998 J. Aud. Eng. Soc., which is a peer-reviewed academic journal [1]. They found an average diffscore (using the same rating system as the MPEG test) for 64 kbps/channel AAC of -0.40, with 24 subjects. This is about the same (not statistically different) as the MPEG test, showing the validity of that test (the two tests used different critical material).
I would have cited this in my original post, but the MPEG test is available online, and you have to go to the library to read the JAES report.
Second, I think you misread the MPEG report -- there are 22 subjects included in the test results. There were 3 *sites* and 3 *listening positions*, but 31 subjects in all were tested. 8 had to be removed in post-hoc tests as detailed in Section 10.3 of the report.
I agree on the relative accuracy of Ref/A/B tests compared to subject-switchable. It's a tradeoff between time and cost and robustness. The Toronto test used hard-disk playback and allowed users to switch.
Finally, I am not trying to deny that both tests show there is a difference between AAC and original recording -- this is very clear. But it's not a conspiracy nor a secret, and AAC is much closer to transparency than any other coder at equivalent bitrates. It depends on the application whether or not the quality degradation you get with AAC is acceptable. It's better than FM radio, cassette tape, and MP3 -- all formats which have proven viable in the marketplace. Thus, I conclude that the coding artefacts in MPEG-4 and AAC are not barriers to their marketplace acceptance.
If you really want something to argue about, you should attack FM radio. Start yelling "the quality of radio is bad, you shouldn't listen to it!" and see what people say.
Although I don't know exactly what you mean by "rival"; I don't think this is a scientific term.
It is true that MPEG-4, like other modern audio formats, is a lossy psychoacoustic coder, and so the bits aren't the same as on a CD. But MPEG and independent groups have done extensive testing to make the sound quality as good as possible.
MPEG found when testing Advanced Audio Coding (AAC), which is the basis for the the high-quality coder in MPEG-4, that at 64 kbps/channel and higher rates it acheives "Indistinguishable Quality" as that term is defined by the European Broadcasting Union. (This means that a certain percentage of highly-trained listeners cannot hear any difference in a formal listening test with certain special methods.)
Anyone who really wants to criticize MPEG-4 sound quality should first read "Report on the MPEG-4 Stereo Verification Tests" by Meares, Watanabe, and Scheirer, available here.
Slashdot Mirror
I'm not sure what 1995 you're thinking about, but I was buying books from Amazon.com in 1995. I have a single account with a continuous order history all the way back.
Iran is not a signatory to WIPO. They do not respect copyrights of works produced in other countries.
http://en.wikipedia.org/wiki/Iran_and_copyright_issues
Just wanted to mention that I *did* post in that thread (well, a closely related one, anyway). Thanks.
Mr. Pibb + Red Vines = CRAZY DELICIOUS
> pick up a non-nerdy book like some good manga or something
/. phrase of the week.
This has to be the
Well, in the 3rd paragraph of Chapter 1, he sets himself right alongside Goedel and Turing. Off to a great start!
There's still a lot of lingering hostility left over from that aquaria debate. For those that weren't tuned in to Usenet the first time around, check out the thread starting here. It's astonishing how worked up gardeners get.
The fact is that most of the things that most people like to use the WWW for--such as e-commerce and Slashdot--cannot be built on Freenet, since it has no cookies and no memory. Given this, I can't see anything happening with Freenet except that it becomes a huge storehouse for illegal porn, pirated MP3s and 3l33t w8r3z.
It's a shame, because the potential political benefits that it raises, by allowing dissident speech in repressive countries, is great.
I guess my question for the Freenet developers would be: I am not a pirate, a privacy nut, a political dissident, or someone trying to spread illegal trade secrets. What does Freenet offer me? And are these benefits broad enough to a broad enough segment of the world population to create the momentum needed for Freenet to work sociologically as well as technically?
-- Eric Scheirer
MIT Media Laboratory
I wrote a long column for MP3.com about the dysfunctions of SDMI; you can read it here. The head of SDMI, Leonardo Chiariglione (who is also the head of MPEG), responded in his inimitable manner here.
So that your program can be interestingly formatted, like this one.
1. It's important to understand the difference between the California case and the 2600 case. The California case is a theft-of-trade-secrets suit, which is unlikely IMHO to succeed. The 2600 case is a suit under the Digital Millenium Copyright Act, which specifically prohibits decryption devices, even when they are constructed through proper reverse engineering. It is much less clear IMHO that DeCSS is not in violation under the DMCA--argument for the defense hinges on the question of whether DeCSS is or is not *necessary*, and has as its *sole purpose*, to acheive interoperability with other DVD players.
2. In some ways lower-court rulings in favor of the DMCA are likely to be of benefit in the long run, because they will accelerate the process of getting higher-court review of the constitutionality of the DMCA. Many legal experts believe that the DMCA is not constitutional. It takes court cases lost in the local jurisdictions in order for this to come out in the legal system. Thus, the most important thing now is to support 2600 and the EFF to continue the fight so that eventually the whole DMCA can be thrown out. Somebody has to be the test case, and it's better if it happens sooner rather than later IMHO. If we won every local case brought under the DMCA due to technicalities, the DMCA and its horrible ramifications would remain in force. Better to lose some small and meaningless fights in order to defeat the DMCA.
This particular fight is about as meaningless as one could be, since there is no practical effect on the Linux or DVD world at large from the ruling. Only the defendants and their contacts are enjoined, so DeCSS distribution is not limited in any important way. Plus I'm sure 2600 is happy for the press coverage.
You can read the DMCA here.
The "reference software" (slow, user-hostile code to demonstrate how the standard is supposed to work) was completed in August and the source is available as part of the spec. Non-MPEG organizations are already building tools for user-friendly use of the standard.
The whole MPEG-4 Audio standard (not including Video or Systems) is about 1200 pages long. It is formally ISO 14496-3:1999 and is divided into 6 Sections:
The part that is most like MP3 is Section 4. Section 4 enables music and other wideband audio coding from 16 kbps to 64 kbps/channel. At the high end, the quality is nearly transparent -- most listeners will not be able to tell the difference between the coded and the original signal. MPEG-4 GA at 96 kbps (stereo) gives about the same quality as MP3 at 192 kbps (stereo) -- thus, files are half as big for the same quality.
There are no "layers" in MPEG-4 Audio.
Some of the sections of the standard (2, 3, 4) are protected by patents and cannot be freely implemented. Section 5 is not protected by patents and can be freely implemented without paying license fees.
Here is the hype from the beginning of Section 1:
ISO/IEC 14496-3 (MPEG-4 Audio) is a new kind of audio standard that integrates many different types of audio coding: natural sound with synthetic sound, low bitrate delivery with high-quality delivery, speech with music, complex soundtracks with simple ones, and traditional content with interactive and virtual-reality content. By standardizing individually sophisticated coding tools as well as a novel, flexible framework for audio synchronization, mixing, and downloaded post-production, the developers of the MPEG-4 Audio standard have created new technology for a new, interactive world of digital audio.
MPEG-4, unlike previous audio standards created by ISO/IEC and other groups, does not target a single application such as real-time telephony or high-quality audio compression. Rather, MPEG-4 Audio is a standard that applies to every application requiring the use of advanced sound compression, synthesis, manipulation, or playback. The subparts that follow specify the state-of-the-art coding tools in several domains; however, MPEG-4 Audio is more than just the sum of its parts. As the tools described here are integrated with the rest of the MPEG-4 standard, exciting new possibilities for object- based audio coding, interactive presentation, dynamic soundtracks, and other sorts of new media, are enabled.
Best regards,
-- Eric Scheirer
Editor, ISO 14496-3 (MPEG-4 Audio)
More info: http://sound.media.mit.edu/mpeg4/audio
MIT issued an official press release praising the hack. You can read it here.
For MPEG-4, the licensing authority details are not yet set, but there is general intent to provide a similar one-stop mechanism. I would expect the legal machinery to be in place within a year. There are a lot more patents that apply to MPEG-4 than to MP3, maybe as many as 25 or more (but see below).
Purely in my own opinion, I think that ISO is not nearly strict enough about managing patent claims on the MPEG standards. The basic argument, which is a good one, is that all of the relevant technology is held by patents, and good standard codecs could not be made if patented technology was not admitted. But IMHO the patent-holders are today allowed to abuse their privileges under this system.
Of course there is a point to standards that are protected by patents -- many companies make a lot of money from them in spite of the licenses they have to pay (MPEG-2 video decoders, for example). But these patents also have the effect of making it difficult for hobbyists to develop conforming freeware and open-source products. This may, in fact, be the goal, but it is wrong IMHO.
Not all of MPEG-4 audio technology is protected by patent. The audio synthesis codec that I developed at MIT and contributed to MPEG has been released into the public domain at my request.
-- Eric Scheirer
Editor, ISO 14496-3 (MPEG-4 Audio standard)
The situation was much different with Java. Sun wanted simultaneously to have Java be an International Standard, but also reserve the right to make unilateral changes to the technology if they decided that it was in their interest to do so.
The difference is clear for MP3 -- no individual company controls the MP3 standard; if revisions need to be made, this is decided by MPEG as a whole, not by Fraunhofer or any other individual company. Sun could not accept letting the other ISO members share in the role of deciding the progression of Java. This is their right, as they own the Java technology, but then the technology can't be an International Standard.
-- Eric Scheirer
Editor, ISO 14496-3 (MPEG-4 Audio)
Eric Scheirer
Editor, ISO 14496-3 (MPEG-4 Audio standard)
And I think the agenda highlighted for H2O is dead-on.
There's a real question today about the breadth of solutions that can be developed in the open-source model as it currently exists. Can "niche" products attract the developer base that they need to leverage the benefits of the bazaar? This is still unproven, I think.
The folks at Harvard that are involved with this are serious heavyweights when it comes to public policy analysis and lobbying strategy. It behooves the community (IMHO) to think about the issues here sooner than later. Some of us aren't fans of our local or national governments, but it surely must be better for the Government to think that open-source is a Good Thing than that it's a Bad Thing.
The promise of convincing governments of this (obviously, USA-centric at first) is the goal of the organization, and I think it's a worth one.
-- Eric
-- Eric
Technical work on MP4 was completed in October 1998, and it is in the publication process at ISO now. I'm not aware of any fully-functional encoders/decoders yet.
-- Eric Scheirer
Editor, ISO 14496-3 (MPEG-4 Audio)
In summary, some of the technology from VQF is included in the MPEG-4 standard, but a .VQF file is not really an MP4 file since MP4 has its own file format, and VQF is only used in certain restrictive ways within MPEG-4.
The story doesn't mention that Platinum is doing this as part of their process of getting rid of all of their VRML assets and staff. They fired a thousand people on Tuesday, including everyone there working on VRML. *Then* they decided to talk about open-sourcing the Cosmo code.
A majority of the people who are/were heading the projet to create a next-generation VRML standard worked for Platinum and are now without employ. It remains to be seen whether this is the end of VRML development, or whether the VRML community can reorganize itself around open-source development.
I would have cited this in my original post, but the MPEG test is available online, and you have to go to the library to read the JAES report.
Second, I think you misread the MPEG report -- there are 22 subjects included in the test results. There were 3 *sites* and 3 *listening positions*, but 31 subjects in all were tested. 8 had to be removed in post-hoc tests as detailed in Section 10.3 of the report.
I agree on the relative accuracy of Ref/A/B tests compared to subject-switchable. It's a tradeoff between time and cost and robustness. The Toronto test used hard-disk playback and allowed users to switch.
Finally, I am not trying to deny that both tests show there is a difference between AAC and original recording -- this is very clear. But it's not a conspiracy nor a secret, and AAC is much closer to transparency than any other coder at equivalent bitrates. It depends on the application whether or not the quality degradation you get with AAC is acceptable. It's better than FM radio, cassette tape, and MP3 -- all formats which have proven viable in the marketplace. Thus, I conclude that the coding artefacts in MPEG-4 and AAC are not barriers to their marketplace acceptance.
If you really want something to argue about, you should attack FM radio. Start yelling "the quality of radio is bad, you shouldn't listen to it!" and see what people say.
All the best,
-- Eric
It is true that MPEG-4, like other modern audio formats, is a lossy psychoacoustic coder, and so the bits aren't the same as on a CD. But MPEG and independent groups have done extensive testing to make the sound quality as good as possible.
MPEG found when testing Advanced Audio Coding (AAC), which is the basis for the the high-quality coder in MPEG-4, that at 64 kbps/channel and higher rates it acheives "Indistinguishable Quality" as that term is defined by the European Broadcasting Union. (This means that a certain percentage of highly-trained listeners cannot hear any difference in a formal listening test with certain special methods.)
Anyone who really wants to criticize MPEG-4 sound quality should first read "Report on the MPEG-4 Stereo Verification Tests" by Meares, Watanabe, and Scheirer, available here.
Best to all,
-- Eric
Editor, ISO/IEC 14496-3 (MPEG-4 Audio)