Ogg Vorbis Gaining Industry Support

An anonymous reader writes "While Ogg Vorbis format has not gained much adoption in music sales and portable players, it is not an unsupported format in the industry. Toy manufacturers (e.g. speaking dolls), voice warning systems, and reactive audio devices exploit Ogg Vorbis for its good quality at small bit-rates. As a sign of this, VLSI Solution Oy has just announced VS1000, the first 16 bits DSP device for playing Ogg Vorbis on low-power and high-volume products. Earlier Ogg Vorbis chips use 32 bits for decoding, which consumes more energy than a 16-bit device does. See the Xiph wiki page for a list of Ogg Vorbis chips."

Informal poll

Ogg Vorbis is:

o An invading species
o The best audio format
o Can be bought at Ikea

Re:Informal poll

An invading species of audio format sold by Ikea!

No wonder it's not used in many audio players!

Run away! Run away!!!! :-P

Re:Informal poll

They're characters in Pratchett books. Okay, they claim that Ogg was from Netrek.

Branding: "Ogg" vs. "Vorbis"

Well, the problem is that you don't understand what "Ogg" and "Vorbis" (and "Theora") actually are. There's actually two different things here: codecs and container formats. "Ogg" refers to the container format; it's comparable to Quicktime, AVI, or Matroska. "Vorbis" and "Theora" refer to codecs (audio and video respectively); Vorbis is comparable to MPEG 1 layer 3 (aka MP3) or Advanced Audio Codec (AAC) and Theora is comparable to MPEG 2, DivX or H.264.

So, when people say "Ogg Vorbis" what they're actually referring to is a Vorbis audio stream inside an Ogg container. Presumably, it's possible to have a file with a raw Vorbis bitstream (without the Ogg container), and it's certainly possible to have an Ogg container without a Vorbis bitstream. This is also why Ogg Theora files have an .ogg extension; they're actually files with a Theora video stream and (probably) Vorbis audio stream, inside an Ogg container.

Re:Branding: "Ogg" vs. "Vorbis"

The real issue is that people use extensions based on the container format, which is totally irrelevant to anything. Why would you ever care that your file uses the Ogg container, but not care what codec it uses or even what sort of media is encoded in it? I give all of my Ogg Vorbis files the extension ".audio", same as my mp3 files. Any software that's likely to be able to play them is going to be able to tell from the file contents what container format it uses. But it's useful to me to know whether I should be playing a file with a music player or a video player.

Of course, I think most people would be more comfortable giving their Ogg Vorbis files the extension ".mp3", since that's commonly and unambiguously used for files containing only audio.

AVI does the same thing.

AVI is a container format; you can have any number of codecs stored within an AVI file. Same thing for WAV.

Why is this a problem for Ogg but not AVI?

Re:Branding: "Ogg" vs. "Vorbis"

Can you name a portable music player (generally referred to as an "mp3 player") that doesn't play mp3s? Even if it isn't superior in every way, it is where it's at. And that's the problem.

Re:Informal poll

Hence, we should call them "ogg" files.

...when we're talking about the file format, that is. In this case, however, we're talking about chips designed specifically to decode the Vorbis audio stream, so "Vorbis" (without Ogg, unless the chip is capable of understanding the container format too) is the appropriate name to use in this thread.

Re:Informal poll

The noob looked like Scorpion with the ninja garb, but completely black.

Actually, he looked more like Sub-Zero with the completely black ninja garb.

MP3 License

If you look at the price list for this chip it states that "Prices include MP3 license of Thomson Multimedia."

Wasn't the point of Ogg Vorbis to have a codec free of licensing?

Re:MP3 License

I think this is the whole reason. If someone is looking for a chip that does Ogg, they can choose this one. If they are looking for a chip that does MP3, they can choose this one.

Business wise, which is better? Selling an MP3 decoder chip for $0.10 each (just a guess), or selling an MP3/Ogg decoder chip for $0.10 each? Since there are no patents, adding Ogg support is free, but adds value. Lots of people may want chips that can play MP3s (GPS, Cell Phones, MP3 players, calculators, EVERYTHING plays MP3s), but how many would buy a chip that only did Ogg? I doubt that market is nearly as large. Added value.

That's my guess. Your product (possibly with a little bit of extra programming) could even use both. MP3 for things you want at a higher quality, Ogg for things less important. Maybe you are upgrading your old product. You can keep all the old samples MP3 and just add the new samples as Ogg. Who knows.

Re:MP3 License

MP3 for things you want at a higher quality, Ogg for things less important.

You've got that backwards. Vorbis is a better codec (in terms of sound quality at a given level of compression) than MP3.

Re:MP3 License

I was wondering about that. I've never used it.

Unless I'm mistaken, just about everything (e.g. Windows Media Audio, AAC, Vorbis) is better than MP3. What's debatable is how the former three compare to each other.

Perhaps Ogg for all internal sounds to a device, and the MP3 capability for sounds the user wants to add so they don't have to use a "weird custom proprietary" format (despite the the fact it's not).

That makes sense, since even if the user has heard of Vorbis he doesn't necessarily want to re-encode (and certainly doesn't want to transcode, as the resulting file would sound worse because the previous encoding to MP3 would have thrown away information that Vorbis would need).

Re:MP3 License

It all depends upon the bitrates and the application. Also mp3 is not as static a target as you might think, with the advancements in lame over time.

If you believe the folks on hydrogen audio, when strong music fidelity is a concern, WMA has unpleasant artifacts at most bitrates, save the very high where even still mp3 is probably your best bet for transparent lossy compression. Well, maybe wavepack if you're really hardcore, but mp3 seems "good enough" for most ears, while wma does not.

At lower bitrates (128kbs down to 40kbps or so) mp3 isn't as competitive, and the winners at different bitrates seem to be AAC and Vorbis AoTuV. This is really impressive for Vorbis because it is a _much_ simpler format, without various special tweaks and features to help out at certain format ranges. The specialized features of AAC help it hit certain windows, but also cost overall in format complexity, which has a minor effect on size overall, and a major effect on implementability. Vorbis by contrast is much simpler and therefore re-implementable, although market forces have not pushed as hard for tuned implementations.

Once you start heading south of 40kbps, you probably aren't really so interested in music anymore, and other more focused audio codecs, probably for speech, are what you'll want to look at.

But the point is mp3 still has some application domains (~200-300kbps, full spectrum music) where it is probably the best format in terms of fidelity and certainly implementatability, primarily because of the maturity of the encoder sourcebase. Surprising, but true.

Personally, for portable music replay, I use Vorbis AoTuV at around 160kbps, because while in testing on my portable player I could often tell the difference, the differences were never offensive. It's possible that some form of aac encoder could achieve this as well for me, but FAAC could not, and I am not willing to pirate and run windows or mac binaries just to encode music in formats that aren't broadly supported anyway on current devices (especially mine). WMA had an unpalatable flat quality at all rates I tested. Maybe it's improved but I was really testing for novelty. That format is even worse than AAC, which at least has an open specification.

OGG is the Game Industry's Favorite Format

OGG Vorbis is used all over the place in the Video Game Industry, since it's free, well documented, sounds great, and has source code available. I think MP3 is only in the forefront of people's minds because the news media coopted the name of that format to encompass all lossy compressed audio schemes, the way "Kleenex" is used by some people to refer to generic facial tissues.

That said, I've used Vorbis playback in an audio library I wrote, and thought it was probably the easiest part of the whole project.

Re:OGG is the Game Industry's Favorite Format

I mentioned ogg to the lead sound programmer at the last games company I worked at and they started using it at their generic format. It still had to be converted to a console specific format for the runtime, because the hardware was designed to handle certain types of streams, and audio isn't cheap to transform cpu-wise. Of course that was ps2/xbox/gc and I'm under the impression that they were able to do a lot more runtime processing of the audio on the "next gen" consoles, but I don't know what role ogg played. Certainly the memory bandwidth savings off ogg in the runtime may outweigh the cpu costs, but again, that's probably something most companies are still working out and I don't know from experience.

Re:OGG is the Game Industry's Favorite Format

Another advantage with ogg over mp3 is that it supports more than 2 channels. The video game industry, especially those doing dev on next-gen consoles, are quite aware of this.

Re:OGG is the Game Industry's Favorite Format

Consoles have hardware-accelerated ADPCM compression for normal sfx, but audio streams can certainly be compressed with Vorbis. All the commercial audio libraries support it at this point, or else it's easy enough to add the support yourself. The next-gen engines have plenty of horsepower to spare for vorbis decoders - it's really not that expensive as long as you don't go too crazy with simultaneous decodes.

Our company is switching from mp3 to vorbis for our upcoming projects - it's definitely a better format for a closed system such as games. As is oft-mentioned here, it's a better-sounding codec at lower bitrates, which is important for MMOs, since occasional updates are expected - and saving bandwidth wherever possible certainly matters. And, it has a few technical benefits such as sample-accurate decoding (MP3 decodes in blocks, so you have to write additional kludges to get around this), which is helpful for loops.

It's nice to hear the format is picking up a bit of steam. I've had my eye on it for a long time, and have been impressed with the steady progress that has been made.

Re:OGG is the Game Industry's Favorite Format

Also, Ogg Vorbis is much more predictable.

Apparently, you can't take apart an MP3 in a deterministic way. That is, if you hand a compressed block to the MP3 decoder, you could get back an uncompressed block of any size, and it's not possible to determine this size ahead of time. You can partially decode blocks ("Decompress in to this buffer up to a maximum of N bytes,"), but then you can't restart the decoder from exactly where you left off. This means you have to either re-decode the entire block and throw away what you've already used, or blindly move on to the next block and hope no one notices the pop. This sort of sloppiness is generally frowned upon in game programming circles.

Vorbis apparently doesn't suffer from these shortcomings. And it sounds better.

This imparted to me by an experienced console game programmer, as relayed through my highly imperfect memory.

Schwab

Re:OGG is the Game Industry's Favorite Format

There's no problem with decoding MP3 in a fixed buffer size. Each frame contains exactly 1152 audio samples.

The MP3 problem you might be thinking of is the bit reservoir: Constant bitrate MP3 only pretends to be constant bitrate. If you look at the spacing between MP3 frame headers it looks like each frame is exactly the same size. But they're really not: frames can borrow bits from nearby frames, so the compressed data at one place in the stream doesn't necessarily decode to the decompressed samples that nominally correspond with that frame. Thus it's tricky to determine where you have to start decoding if you want to seek to a given sample number, and the naive seeking method could be off by about +/- 0.25 seconds.

That problem is specific to MP3; I don't know of any other audio format that suffers from it. All Vorbis had to do to fix it was be logical and put each bit in the frame it's supposed to be in, not in some random other frame.

Storage vs processing vs quality

There will always be some sort of trade off between cost effectiveness of storage vs processing and cost effectiveness. There are no obvious winners, and the best solution will change as the memory vs micro prices change.

Many voice mail systems only use 32kbps sampling and achieve fine results for that purpose, and the algorithms are easy enough to render on a 8-bit micro costing 50c.

When it comes to medium quality sound then there are basically two routes you can take: 8 bit micro (or even some dumb logic)running less fancy algorithms and a bit more flash/rom to store more verbose sound data; or more compressed sound and a flashier micro to run a heavier algorithm. You can now get 32-bit ARM micros for less than $1 making the second option reasonably feasible at low cost.

However flash is very cheap. NAND flash only costs approx 2c per MB (for multi-MB chips, so small chips are going to cost more per MB). You can fit a lot of "mama" phrases in a couple of MB. As a result you don't want to spend too much money on micros to save on flash.

money talks

Ogg Vorbis is gaining popularity mostly because of the price per unit. When you make millions of dolls a year and you have to pay a $0.10 licensing fee per unit if it plays voice prompts in MP3 format, that starts to get pretty expensive. If WMA, AAC, MP3, or any other codec was cheaper and did not require significanly more flash memory to store, they'd be using that instead.

Re:money talks

It is no longer a dime anymore. The IC prices are not listed online, but the per device prices are for hardware items.

http://www.mp3licensing.com/royalty/hardware.html [mp3licensing.com]

At the bottom of the page is tha item that unless you buy chips with the license, the minimum for doing it yourself is $15,000 USD. If you are making a limited quanity of an item, the minimum can be a showstopper unless you buy chips from someone else, which may also be a little expensive. Dropping MP3 can save a chunk of change since a free alternative exists.

It's the PNG/GIF thing all over again.

No, Worse because M$ Squished it.

It's the PNG/GIF thing all over again.

Except in this case M$ gave music player makers a choice: our way or the highway. The Janus DRM license actually forbade the use of ogg. Though this was shot down by the EU [theregister.co.uk], you might imagine the pressure is still there. Well, it was until M$ hosed every one of them over by dumping the former "Plays for Sure" for whatever their new "service" is. You would think they would revolt given they can't win in the M$ world.

Openness == Interoperability

This highlights the reasoning that large corporations such as IBM, Novell, and Sun have adopted open source methods: it lowers their bottom line. They pay programmers to work on open-source projects and they more than recoup the costs through savings in other areas such as interoperability. Open-source breeds open-standards and when basic infrastructure such as audio support is basically "free" then costs are lowered more by using common-infrastructure between manufacturers vs. constantly reinventing-the-wheel or developing your own library of common code/components. Reiterating simply, it's cheaper to pay programmers to develop free infrastructure code and give it away to reap higher profits from reduced costs in other areas such as interoperability.