Pacific Connection(英語)

Net Music: MP3 and its Cousins Are Shaking Up the Recording Business

It has been said that "information wants to be free"--or to put it more cynically--piracy is an inevitable part of human nature. That precept is being put to the test by MP3, an audio file format that approaches the quality of audio CDs. For musicians, recording labels, and their public, MP3 is both an opportunity and a temptation. The format has already opened a new "channel of distribution" for recording artists that bypasses the conventional brick-and-morter shop, permitting musicians to reach new audiences directly, around the globe. The temptation is piracy, pure and simple. MP3 is an open standard with no built-in access restrictions. If I have an MP3 file of a cut by best sellers Metal Essentials or the Goo Goo Dolls, I can put it on my website for the world to download, at least until I'm caught.

MP3 is at this point the de facto standard compression/decompression algorithm--or "codec." The "MP" in MP3 comes from MPEG, the Motion Picture Experts Group. The "3" refers to the Audio Level 3, the digital compression component of the MPEG standard. As MPEG is to video and JPEG is to static images, MP3 is to high quality audio, a way to compress CD-quality sound, typically by a factor of 10 so that 1MB of storage can hold a minute of music. For the listener, the result is not quite that of the original, but it's close. Playback only requires a multimedia PC and a free MP3 player program such as WinAMP. You can also download MP3 files to a portable device and listen to them on the subway.

One downside to the format is storage. RAM storage on portable devices devices is typically limited to about an hour, although you can add more capacity through additional flash RAM cartridges. On a hard drive, 2 gigabytes will store about 60 CDs. Not bad, although a single disk crash could blow you're whole collection. Users can create their own MP3 files using packages that are called "encoders and rippers." In the music industry, a ripper is the software that records a music file from another format, such as a CD or digital audio tape. The encoder then converts that file back into MP3. Serious collectors could also use a CD-R drive to burn their own CDs--opening the door to MP3 formatted music being played in a car radio.

While MP3 clearly has a headstart, it is not the only format for net music. Microsoft is in the game with MS Audio 4.0. The format will include a copyright clearance facility that will check a website that could impose additional conditions, including payment, before the file can be opened. But some recording executives are concerned that Microsoft in the past has supported MP3 and is bridling at what it sees as enforced standards.

Meanwhile, Real Networks, Inc. --known for its Real Player streaming video and audio software--has been very busy. The company paid up to $75 million in stock to purchase closely held Xing Technology Corp., which among other things, makes MP3 ripper and encoder software. The week before, Real Networks aligned itself with IBM. The deal called for Real Networks to develop its own audio player based on IBM's Electronic Music Management System. And the company announced a product called Real Jukebox, which will record and play audio in MP3 and its own G2 format, among others. The software is controversial because it will allow users to quickly copy and e-mail one or more tracks from a music CD, albeit with a copyright warning.

With these and other codecs competing with MP3, the question remains: can a lockable format succeed? Some say it must if music over the Internet is to become a true commercial venture. Others say it won't because people won't pay for what they can steal--or that it can't, because "secure" audio files are just too easy to break.

The argument for enforcing an artist and label's copyright privileges would seem to dwell right side of the angels. We all do what we do in life with mixed hopes of love and money. Just as software developers hope to see a reward for their lines of code, so do recording artists and composers, struggling or successful, hope to see their art sustain them. The rest of the entourage--managers, record label producers, sound engineers--also need to feed their families. Copyright protection, it could be argued, is at the very core of what it means to be a professional, a way of ensuring that your talent is a sustainable profession. Besides, copyright infringement is against the law, with hefty fines and even jail sentences imposed on its biggest offenders.

But the history of the software industry suggests the story is more complicated than that. In the early days of off-the-shelf software, vendors encrypted the disks in an effort to prevent unauthorized duplication. That practice inevitably led to "skeliton key" software to unlock the devices. The two technologies escalated in Darwinean fashion much the way virus and anti-virus software does today. Ultimately, customers voted with their pocket books, with convenience trumping copyright enforcement.

The open source movement also illustrate's the potential and preference for free software. While online music files are not a collaborative medium, the Internet community can repay the favor of free music with a cult following, which can later translate into people buying tickets when your gig comes to town. At least, that's the theory. And then there's the case Divx, a format promoted in the U.S. that allows customers to purchase a DVD dics for a single viewing at a low cost. If you would like to convert a disc to unlimited viewing, you pay more. A pragmatic approach, perhaps, but sales have been poor. So if recent history is any indicticator, the very fact that MP3 is "unlockable" gives it an advantage over other high-end sound formats in terms of adoption. And that fact that it's open--your for the taking--makes it more attractive still.

Music from a virtual music store

The quick, cheap propigaton of MP3 files is making the recording industry very nervous--as change often does. Right now, recordings are promoted through personal appearances and radio exposure. If you want a CD, you go down to your local CD shop to buy it. By contrast, MP3 permits even the smallest publisher to pre-release a couple of songs and get them in the hands and ears of the public. Indeed, some publishers have uploading musical content to the Internet even before they deliver it to the stores. Last March, rocker Tom Petty made his new single available on and got 150,000 downloads in just two days. His label, Warner Bros., was not happy.

Other musical offerings on are less well known, but certainly wide ranging. Categories range from hip hop and jazz to African, oceanic, and classical. Click on the Japanese section, and you can access more than 150 songs from performers like Electric God, Takehisa Takatsuki, and Powered Express. You can also click in regionally by city to get music from Nagoya, Sasebo, or Setagaya among others--althouth the cities outside of Tokyo have only a few offerings.

Of course, you can also publish music on your own website. Glen Richards of Houston, Texas translated his collection of 1920's vintage 78 rpm jazz records to the MP3 format and upleaded them to his own web. The legality is questionable: Richards himself admits the copyright issue is "definitely a grey area." But the response was clear. Richards did nothing to promote his offering, and yet in five months, 4,000 people from as far away as Yugoslavia, Turkey, Malaysia and New Zealand had paid his site a visit.

And while some have called the typical MP3 listener a "insomniac college student," the audience is clearly much broader, perhaps shifted toward more conventional tastes by a burgeioning demand from office workers looking for relief from the tedium of work. Michael Robinson, the CEO of the site, told The New York Times that less than five percent of the music downloaded from his site is requested by people with college e-mail addresses. The majority seem to come from people in cubicles looking for music during lunch. The trend suggests that downloadable music may not so much be displacing CDs, as growing the market for music in general by enabling people sample new recordings during their lunch hour. In the U.S., the real losers may be radio stations, which advertise at night to attract people to listen to them from work by day.

As evidence of the mid-brow values of MP3 listeners, the most popular file request access from earlier this year was Beethoven's Moonlight Sonata, performed by pianist Richard Morris of Cincinnati. Last March, that piece alone got 100,000 "hits," a substantial number for a classical recording. And the company once known as GoodNoise, which sells music online, acquired in part for its more broadly appealing name. Still, rock reigns supreme. The site recently offered tracks from Frank Zappa--like 20 Small Cigars and Carolina Hard-Core Ecstacy--for 99 cents each.

Violated copyrights

Some believe the recording industry's hostile reaction to MP3 stems largely from the format's potential to upset the status quo. But let's face it, MP3 files spawned over the Internet represent a temptation for piracy that many will find hard to resist. An illegally uploaded MP3 file is yours for the taking, whether you live in Tokyo, Texas or Timbuktu. And unlike an illeagle CD duplication factory, the perpetrators are difficult to track. Shut down one site, and the purloined music can appear in anther.

The Record Industries Association of America (RIAA) got off to a rocky start on the MP3 issue by going after Diamond Media in an unsuccessful stop to stop Diamond from selling its Rio PMP300 digital player. The RIAA argued that the Rio violated the American Home Recording Act, which was instituted to ensure that digital audio tapes (DATs) aren't used to make multiple copies of a digital original. Diamond countered that the Rio is simply a computer peripheral, and that PCs are exempt from the act. Moreover, the Rio doesn't make "recordings" in the conventional sense. So far, the Diamond has prevailed. Selling for less than $200 and powered by a single AA battery, the device stores music on 32 MB of flash memory.

Indeed, in the Amercian tradition, Diamond shot back with a countersuit, asking for treble and punitive damage for "intentional misconduct aimed at injuring Diamond." Diamond's lawyer, Andrew Bridges, said that the countersuit is a response to "a conspiracy between RIAA and others to restrain trade and restrict competition among manufacturers in portable MP3 devices." He said that Diamond Multimedia also incorporated a serial Copy Management System into Rio, and that he believes the RIAA's "real goal is stopping the legitmate MP3 market." (Ironically, the RIAA has dubbed its new certification level for best selling recording artist as "the Diamond Award.")

Diamond has an allly long-time industry pundit John C. Dvorak, who wrote that the rampant MP3 bootlegging is the record industry's own fault. Dvorak complains "nowadays a person usually has to shell out $16.95 for an entire CD, which seldom contains more than two decent songs." MP3, he says, fills the gap once occupied by single recordings, in which you got a few good songs for cheap. Dvorak also points out that while CDs cost less to manufacture than tape cassettes, they cost significantly more at retail, and that the recordign industry enforces the high cost by threatening to pull co-op advertising from record stores that offer discounts. "I sure don't condone illegal activity but I'm not surprised. This is a trend that a greedy cavalier industry brought on itself."

Undaunted, the RIAA surfs the web on its own, claiming that it shuts down hundreds of illegal sites each week. While few have been punished, one company that compressed every Beatles album into a single MP3 file is facing criminal charges. The association says while cassette piracy has dropped by more than 80 percent over the last five years, "more than 100 million computers worldwide are connected to the Internet, and most of them can download and store pirated copies of recorded music." The RIAA has a harder time actually documenting the extent of priacy via MP3, admitting it has only anecdotal information of losses. Instead, the organization simply claims that "the Internet culture of unlicensed use means that theft of intellectual property is rampant, and the music business and its artists are the biggest victims."

Last January, major industry members met to discuss the Secure Digital Music Initiative, a means of enforcing copyrights while insuring higher fidelity. But companies aren't waiting. At this writing, Sony Music Entertainment just announced it would put sell the online equivilant of singles beginning this summer using Windows Media 4.0. That followed a similar move from Universal Music, which said it would work with Intertrust Technologies Corporation of Sunnyvale, Calif. to develop a secure format. "The demand is there, and demand is being filled now by independent labels and illegal content," said Larry Kenswil, who heads electronic commerce at Universal, in a New York Times interview. "It's crazy for us to not recognize demand and move."

That urgency contrasts with the ex-members of the Greatful Dead, the legendary San Francisco group that disbanded in 1995 after the death of Jerry Garcia. In keeping with its 1960s roots, the band has all but encouraged bootleg recordings of its concerts, feeling that the recorded music only boosted ticket sales. Now, the former band's organization has told Dead Heads that the many recordings already on the Internet are theirs to propigate. As Garcia once was reported to have said about the band's music: "When we're done with it, you can have it."

Some Clear Talk About Compressed Audio: An interview with David E. Weekly (

At the age of 20, David E. Weekly hasn't yet graduated from the computer science department at Stanford University. But he's already one of the more influential and savvy observers of audio formats. Weekly was a founding member of the MP3 Audio Consortium, an alliance of webmasters interested in the technology. When he got too busy to maintain that own site, he shut it down, and replaced it.

How do MP3 and its progeny work?
Let's start at the beginning. When we speak, we are compressing air waves. If you hold up a microphone, the waves will cause the diaphragm in the microphone to wiggle, just like the inside of our ear wiggles when sound hits it. This microphone will transmit an electrical voltage corresponding to how squished in the diaphragm gets.
The next step is to take that analog signal and to digitize it, which means quantifying the signal. For CD-quality recordings, a number is assigned between zero and 216. At this point we have a stream of bits that defines how squished in the diagram was at any given point in time, for a whole bunch of different points. I can feed these bits to another computer, which reconstructs these pulsations on a speaker cone and voila, you get playback. That's a full uncompressed digital audio system.
Now the problem becomes that you are sending a whole lot of bits to get the sound to sound good. Look at the math. If you are using 16 bits for each sample, and you have a left and a right channel for normal stereo recordings, and you are sampling 44,000 times a second, that adds up very fast. It amounts to 160 kilobytes for every second of recorded audio that you record. Twelve megabytes for each minute of sound is hardly something that could stream over a modem.
So people have thought about compressing it. You can't use something like PKZip to do that because audio doesn't compress well in that way. Even the most advanced, completely lossless audio compression algorithms can only get you down about 70 percent of what the original file size was. So in order to trim things back to be able to stream over a modem with much less data, we need to be willing to throw some data away. The key is to throw away the data that is least important to the human ear.
Is this similar to what JPEG does?
It's the audio equivalent. The human ear puts frequencies into 25 distinct ranges, called 'critical bands.' And these bands can mask each other. When you are driving, for example, you have to turn the sound up, because the noise of the car masks the music. But that sound is way too loud when you are stopped. The compression algorithm takes advantage of this by looking for recorded signals that can be masked out without greatly affecting the quality of the music. A computer could notice the difference, but you won't be able to hear them.
Is that the basic idea?
Yes--you exploit masking and you throw away the parts you can't hear. As you compress further to accommodate modem rates, you have to throw away bits of stuff you can hear. But masking still gives you some rough idea of priority. First to go are the ones you can't hear, then the ones that probably you won't hear. If you need to crunch this to a really small file size, we'll throw away more bits, even if you are able to hear them.
Is that why they say that MP3 is near CD quality but not exact?
That's not necessarily correct. The more bits you give to MP3, the more accurate it is. If people are listening to recordings in which a lot of the bits were thrown away, it's not going to sound as good as the original.
So MP3 varies in the amount of compression it will do?
Yes, it's a scalable codec. This is true of any of the modern codecs out there.
But it's generally true that MP3 files posted to the Internet deliver near CD quality sound?
Yes. Most people encode MP3 files at 128,000 bps for two channels, so that's 64,000 bps per channel.
Some codecs are claiming to be better than MP3. Is that generally true?
It's probably true. I can say to the MP3 algorithm: given so many bits--make as good sounding a recording as you can. And I ask Microsoft's codec, for example, to do as good a job as you can with the same number of bits. When you listen, you'll find there are a number codecs are superior to MP3.
You've done some sophisticated testing. Can the human ear tell the difference?
It's entirely inappropriate to have a computer program state that a file sounds bad. Subjective criticism is best left up to people. So I wrote my subjective analysis of how I thought the codec sounded, and I leave all of the samples on my website to let other people disagree, and make their own subjective conclusions.
In your opinion, which is the best?
It depends on how much you need to squish your data. If you only need to squish it a bit--if you've got a very big hard drive--use high bit rate encoded MP3s. Nothing I've found sounds as good--it is exactly CD quality. I made an encoding of [Sting's] The Hounds of Winter and it is beautifully crisp. I used many more bits--about 50 percent more bits than most of the MP3 files that you'll find on the Internet--a megabyte and a half of storage per minute.
The next best in the high to mid-rate Microsoft Audio encoded format. I think RealAudio is best for the lower modem bitrates.
What about the security issue?
For practical purposes, none of them are truly secure. All of these sound playing programs work by doing their fancy decryption, of the music, then sending the music out to the sound card. It's a trivial technical matter to write a small piece of software that "pretends" it's a sound card driver, and which in turn writes to a new WAV file. Then you just take the WAV file, compress it and send it to all of your friends.
So no format is actually secure?
Yes, any audio security format is inherently breakable. This is the major thing that people have been overlooking. Right now the programs to do it are not very user friendly. But if people start trying to push a really secure format, and a lot of cool music gets on it, this stuff is going to be so to use--1-2-3 click. And that should not be surprising. This is the lesson that the software industry learned 15 years ago. Because some locked software was difficult to crack, inspired some people to go and crack it. I submit to you that "cracking" any form of music encryption is trivial and that it doesn't even take a sophisticated hacker to do it.
Given that, do you think that MP3 will continue to dominate?
MP3 is not the best format technologically. It was developed about 10 years ago and a number of formats have risen up. But yes, it will continue to dominate--because it is open. In MP3, the algorithm--the method by which one makes the software to play back the MP3 files--is public knowledge. It's out there.
And nobody is trying to license it. You're not a pirate for using it.
Right. Whereas the other formats, like RealAudio, Liquid Audio, and Microsoft MS Audio, are all closed and proprietary.
If a large recording firm starts to support one of those standards, thereby getting highly desirable content onto a non-MP3 format, do you think that will matter?
It will encourage people to use that format. I won't claim that MP3 will be the long term win. But it is the mid-term win because the installed user base is enormous and it's really cool and makes sharing very easy. There's a very large and rapidly developing hardware market around MP3.