Ogg Vorbis—Open, Free Audio—Set Your Media Free

Audio has become one of the killer apps of the network. With the distribution power that the global network offers, the music industry is being reshaped forever.

The boom of audio applications and files on the Internet is responsible for much current litigation surrounding copyright law and music licensing. The record industry is just now figuring out what most early users knew the first time they played an audio file on their computer: it's a new world for artists, listeners and record labels.

At the center of the upheaval are the technologies that make it all possible and a new technology, Ogg Vorbis, is ready to put this revolution into an even higher gear.

Ogg Vorbis is an open-source and patent-free audio codec that is being developed by Xiphophorus along with several other multimedia projects (cdparanoia and Icecast, to name two). Xiphophorus is a collection of open-source, multimedia-related projects and programmers who are working to ensure that Internet multimedia standards reside in the public domain where they belong. The work on Ogg Vorbis is currently funded by iCAST, the entertainment arm of CMGI.

Ogg Vorbis is an open standard, and this is important for a number of reasons. There are few truly open standards in the realm of digital audio. Look at Windows Media, Quicktime or RealAudio. These standards are all closed and proprietary, and because of this, none of the standards interoperate well (or at all outside of their corporate walls) with one another. When was the last time that you could play Quicktime 4 in RealPlayer or vice versa? When will Linux have Quicktime or Windows Media support? Linux and the Internet are founded on open standards, and as multimedia on the Internet and on Linux rapidly matures, the need for multimedia applications like Ogg Vorbis grows rapidly as well.

There are two parts to Ogg Vorbis: Ogg and Vorbis. Ogg is a wrapper format, similar in some ways to Apple's Quicktime or Microsoft's Active Streaming Format. It helps you collect a group of things that belong together. For example, if you have an Ogg movie file, it might contain a Vorbis stream alongside a video stream in another codec. Or the Ogg movie file might contain ten Vorbis streams, one for each language available.

Vorbis is a codec that is written inside the Ogg framework. It is a general-purpose audio codec that is suitable for compressing most audio sources with good results. It doesn't use subbanding like some codecs do, but it does use vector quantization similar to others.

Vorbis is the only codec we've written so far, but not the only one we plan to write. There also are Squish and Tarkin.

Squish is a lossless audio codec, meaning that there is no loss in quality at all, and in fact, the decoded stream would be byte-for-byte identical with the original stream. You might want to use this for archiving master copies.

Tarkin is our fledgling video codec. It's a work in progress, but I can tell you it's based on wavelets, not on the MDCT like most modern codecs including MPEG-4 and JPEG. We're still playing around with it, but it's quite promising.

Codecs are hard to develop. They take a lot of math skills and a lot of time. Once you finish development, you still have to tune it, fix bugs and think of cool new things to add. This is why Ogg Vorbis focuses primarily on Vorbis and the Ogg framework at this point.

A lot of readers are probably wondering why we'd bother to develop Ogg Vorbis with MP3 already enjoying such widespread use. What's wrong with MP3? It's free, right? Wrong.

Have you ever noticed the amazing lack of free MP3 encoders, especially considering how popular MP3 has become? I can count them all on one hand. Some people will remember the famous letter from Fraunhofer back in late 1997. The letter asked for all the open-source and free MP3 encoders to cease and desist or start paying patent royalties. There are around 12 patents on the algorithms used by MP3, and all of them are heavily enforced by the owner Fraunhofer.

This patent enforcement has several negative effects. It's nearly impossible to have a free MP3 encoder because of the licensing fees for doing so. It costs $2.50 per download ($5 if you use the Fraunhofer code). Most of the free encoders disappeared without a way to pay this kind of tribute. MusicMatch, which makes a popular Windows encoder, sold a significant percentage of its company to Fraunhofer in exchange for an unlimited license.

Fraunhofer can change their rules at anytime, too. Prior to 1997, distributing MP3 encoders was fine. Right now, broadcasting in the MP3 format is free, but Fraunhofer stated that he intends to charge licensing fees for such use at the end of this year.

The deals the RIAA cuts for the broadcasting of commercial music are typically one-third to one-half a penny per song, which is quite reasonable considering that Fraunhofer may want to charge you 1% of revenue with a minimum of a full penny per song (these are my extrapolations from the current fees on commercial MP3 downloads). Is MP3 really worth three times more than the music it delivers?

It costs $.50 a copy to license a decoder. These aren't the only costs associated with MP3, and really, some are just my speculations (hopefully the real fee for broadcasting will be considerably lower), but the patent holders can set or change the licensing fees to whatever they want, anytime they want. And, they already stated that they intend to do so at the end of this year for broadcasting. The point isn't whether it's $15,000 or $5. The point is they have the right to set the price however they see fit.

MP3 is an old technology. Audiophiles and programmers have been tuning encoders for a long time, but the technology is not improving anymore. Even LAME, one of the best MP3 encoders around, has new options that break the specification to try to squeeze more quality out. There just isn't anymore room in the format for new tweaks or improvements.

The alternatives aren't great either. Advanced Audio Coding (AAC), which is a part of MPEG-4, has quite a bit more IP restriction than MP3. There's more than one company involved in most of the technologies, which makes licensing even more cumbersome. The VQF format is locked up tightly by NTT and Yamaha. RealNetworks and Microsoft aren't known for their open standards either. Several derivative codecs like MP+ are problematic because they face the same patent restrictions that the regular MP3 codec has.

With all of these inherent problems and the need for a better way to work with audio on the Internet, it's not surprising that a solution would come from the Open-Source Community.