Interview: Ogg Vorbis Frees the Music

While industry giants battle over a secure audio platform, a team of developers is taking a radically different approach -- establishing a completely open source music format free of any licensing restrictions. While still in its development phase, Ogg Vorbis is already giving new proprietary formats such as mp3PRO and Windows Media a run for their money. BetaNews recently spoke with Vorbis Product Manager, Jack Moffitt, to discuss the project and learn more about the recent release of RC2 on the road to 1.0.

BetaNews: RC1 brought some major changes to the [Ogg Vorbis] format and boasted some big numbers to compete with mp3PRO. What is new in RC2 and what do you think the impact of those changes will be?

Jack Moffitt: RC1 was decoder focused, and RC2 is encoder focused. RC1 put all the features in the decoder that were left for 1.0, and RC2 starts to use these features for even better compression and quality. The biggest change is that we're now doing channel coupling, which means using information among multiple channels to compress even more. This release should convince all the people who have been skeptics before. Ogg Vorbis just sounds amazing.

BetaNews: How does RC2 compare with the newly released mp3PRO format?

Jack Moffitt: I'm going to point you to recent posts on the Vorbis mailing lists.
You can do a quick search for 'mp3pro' at http://www.xiph.org/archives/. Search in the vorbis and vorbis-dev lists.

It's very subjective testing things, and of course, I haven't seen a formal test yet. The general consensus seems to be that the recreation of the high end (even based on good guesswork) is not the best idea in the world. Not to mention, they are still locked into a format that can't evolve much. It takes some room to encode the extra data, so an mp3PRO file at the same bit rate on a non-mp3PRO capable player is probably not going to sound as good.

The list seems to think channel coupling will get us past the mp3PRO quality level, and I bet there are a few people who would say we already are. Again, listening tests need to be done, but we certainly have a lot more room to play with quality than anyone trying to make something backwards compatible with mp3 does.

Try it for yourself :)

BN: Explain to our readers, if you could, how Channel Coupling works and the significance of the addition in RC2.

JM: The definitive source for channel coupling for the curious is Monty's stereo document which is in the vorbis/doc directory. Basically there are 8 coupling modes, ranging from 4 and 8 point phase stereo to lossless coupling. While these features are similar in intent to the joint stereo everyone is familiar with, they also work on 3, 4 or however many audio channels you want. If you're very sensitive to stereo separation, or want archival quality, you can do lossless coupling, and if you want to encode aggressively, do things like 4-point phase stereo. Monty explains this all very well in his document, but the end result for users is much smaller files at the same quality.

BN: When can we expect to see 1.0? Is this the last RC?

JM: There will be several more (and probably released more often) tuning/bug fix releases. This is probably the last major change to Vorbis, and everything from this point forward to 1.0 should be making tweaks for quality or fixing bugs that various people have found.

BN: How do you intend to compete with MP3 when so many other new "advanced" formats have failed? Are you looking to gain support from the artists, record companies, computer users -- all three?

JM: A lot of people don't think it's possible; certain there are a lot of
failed formats.

But, Vorbis is a lot different. It's not a proprietary format. It's not tied to the bottom line of some company. No other format has really been able to say that in this space.

Also, it's not hard to compete in a lot of cases. MP3 is fairly expensive, and only getting more so (did you see the new streaming
royalties?). Vorbis is for all intents and purposes, absolutely free. More than a few game companies have been on the lists, e-mailing us
personally, and shipping Vorbis code. Free means a lot in those cases.

Also, it sounds better, and noticeably so for most people. Add to that that it's supported in most of the 'mp3' tools, and it's an easy decision for someone who's tried both. You lose nothing by using Vorbis, but you gain better sound quality, or smaller file sizes.

Hardware manufactures are interested, but have held back a bit (with
exception for Iobjects/Iomega) because of the 'beta' moniker. If we could go back and call Beta1 Vorbis 1.0, we would have got in a lot more places. But we're not going to call something 1.0 unless it's ready.

As for consumers, we're starting to see adoption there. The (unrelated) Ogg Vorbis OpenNap network sprung up to everyone's surprise, and is promoting the vorbis format on file sharing networks. While not everything there is Vorbis, there is Vorbis content there that is being returned in searches and getting it in front of people.

Artists seem to really like vorbis as well. There has been a lot of
interest in getting music on the music page at vorbis.com, which we're
happy to oblige, time permitting. A lot of them are really excited that their Internet distributed songs don't sound so bad anymore. And of course, if they want to sell Vorbis files, they can do so, royalty-free. And I believe at least one site is selling Ogg files for this reason (lawsonmedia.com, I think. It's linked from vorbis.com).

So our strategy to date has really focused on technology vendors. We push it to game companies, library makers, the ripper guys, the jukebox guys, the players especially. Once Vorbis is in all the tools (which it
basically is now to a large degree), users have no reason not to switch, as Vorbis is fairly ubiquitous, and they don't have to worry about their files not working, like with WMA, Real, AAC, mp3PRO, or VQF. No other format has really achieved this, and we are almost there. The Windows Media Player plug-in will be finished soon, and RealPlayer plug-in is almost done (but working), and QuickTime plug-ins are coming along
nicely. At that point, I don't think there will be a worthwhile player that can't play Vorbis.

Sorry for the long answer :)

BN: I'm sure you have seen the recent coverage by the Washington Post comparing formats using members of the Symphonic Orchestra. What are your opinions on the results (WMA beating mp3PRO with OGG trailing behind)? Does RC2 fix some of the concerns people had with the format?

JM: The tests were badly implemented, not being doubly blind, etc. They were anything but scientific. They even talked about stereo image loss which seems quite unlikely since they were testing versions that used true stereo, since coupling was a new feature in RC2. There are a lot of good ears giving us feedback on quality, and quality is our primary concern. No matter what you believe, you should want until 1.0 to make your final judgment, and always keep an eye on the project, since each new release brings big improvements. We fix bugs and artifacts a lot faster than our competitors.

BN: With all of this work being put into a "secure" format, is Vorbis a step back from these initiatives?

JM: No. Vorbis is a step forward. Soon those 'secure format' dorks will
realize this and switch. Why put so much effort behind a concept that the public DOESN'T WANT and HASN'T ASKED FOR. We've been doing fine
with MP3s being open, and Vorbis being more open will be even better. I'm guessing that eMusic sold more music online than all the major labels with their secure formats combined. I have no data to back that up, but I think it's probably true. Last I asked Ted Cohen of EMI, they had 200 albums online. Absolutely pathetic. I have 200 albums in my office, and another 200 in the living room.

BN: I'm sure your team must be extreme excited about the release of RC2, how have you viewed the acceptance of Ogg Vorbis - have people begun using the format?

JM: It's been accelerating always. We're always getting e-mail from artists putting their music online in Ogg format, or from game developers who are switching their sound engines to Ogg. The end user acceptance has also started to gain momentum with the Beta4 and RC1 releases, and I expect that a lot of people will be impressed with RC2.

BN: I have noticed some hardware devices have begun supporting Ogg. Is this a trend we can expect to see on additional devices in the future?

JM: I think so. I just got an e-mail from a hardware vendor today out of the blue who is shipping me a new toy to play with. It took MP3 years to get into hardware, and WMA a long time too. We're doing better than either of those timelines so far. It just takes a little time for it to propagate.

BN: Will upcoming versions of software (Winamp, Sonique, Freeamp, etc.) ship with Ogg Vorbis support, or will the format always require a downloaded plug-in?

JM: Sonique already ships with Vorbis. Freeamp already ships with Vorbis.
Winamp will ship with Vorbis very soon (their official plug-in is already available on the Web site). I suppose we might see it in the next
version of Winamp, if we're lucky. It will get in there though.

Real Networks and I have been working on a Vorbis plug-in for RealPlayer. They have a pretty cut and dry process, using 3rd party QA facilities, that will let us put vorbis on the auto-update server. Microsoft is likely to push back the most, but we'll give it a whirl.

We'll do our best to convince Apple that they should make our plug-ins easy to get for their customers as well. We've had several discussions with them already.

BN: A lot of open source initiatives have failed due to lack of money and support (Eazel, etc.). Because Ogg Vorbis is completely free do you foresee problems down the road with developers willing to spend time and energy to advance the format?

JM: We're not in this to make money, Eazel was. That's a big difference. A
lot of Vorbis was done not-fulltime, and the collaborative efforts helps offload listening tests and QA.

That being said, Monty and I do make money consulting on Vorbis related projects (or Icecast or cdparanoia related ones - those are two other
xiph.org projects). It's enough to live on.

We've started a non-profit. We're not a .com that will flounder. We're a non-profit company developing standards-based patent-free media technologies to make the Internet (and the world) a better place. Plenty of non-profits are engaged in similar activities, although not
often in the Internet space.

We hope to acquire some money from donations and sponsorships, as other non-profits do. We're also looking for corporate sponsorship to some degree, as we'd like to bring more fulltime brains into the mix, especially to work on Tarkin (the video part of Ogg).

We do provide value, even if we aren't in the business of charging people for it.

BN: How many people and/or companies are involved in the development of Ogg Vorbis?

JM: As for people, it's quite a bit. There are probably a dozen active
people now, with about a half dozen regularly writing code or bug fixes. There are many, many people from the community who've helped, including: packagers, bug finders, users who've spread the word, developers adding Vorbis to their own software, etc.

BN: And last but not least, any news for our readers on Icecast 2.0 and the streaming side of things?

JM: USENIX and Dr. Dobbs just did a broadcast of Edward Felten's paper and talk on the SDMI watermarks using Icecast 2.0, so it's starting to get used more and more. We are hoping to have Icecast 2.0 ready at the same time Ogg Vorbis 1.0 is ready or shortly after. We'll see how that goes :)