Vorbis
Vorbis is a
free and
open source lossy audio compression codec project headed by the
Xiph.org Foundation. It is most commonly used in conjunction with the
Ogg container and is then called
Ogg Vorbis. Although the Vorbis format is often simply referred to as
Ogg, this is technically incorrect as Ogg, like
AVI or
Matroska, is a
container format while Vorbis is an audio codec.
Vorbis development began following a September
1998 letter from
Fraunhofer Gesellschaft announcing plans to charge licensing fees for the
MP3 format. Soon after, founder
Christopher "Monty" Montgomery commenced work on the project and was assisted by a growing collection of other developers. They continued refining the
source code until a stable version 1.0 of the codec was released on
July 19,
2002.
The latest official version is 1.1.2 released on
2005-11-28, but there are some fine-tuned
forks available, such as
aoTuV beta 4.51. Source code (called
libvorbis) for the Xiph.org release is available from the official
Vorbis.com web site, while many Windows binaries can be downloaded at
Rarewares.org. Source code and binaries for the aoTuV release are available at the author's
own site.
The Ogg Vorbis combination has proved popular among
open source communities. They argue that its higher fidelity and completely free nature make it an excellent replacement for patented formats like
MP3. However, MP3 has been widely used since the mid-
1990s and
as of 2006, remains the
de-facto standard in the consumer electronics industry. Of the many consumer products which support lossy compressed digital audio, virtually all support playback of MP3 audio. MP3's overwhelming popularity can also be seen in
peer-to-peer networks. A 2005 study found 64% of audio files on P2P networks to be encoded as MP3, while Ogg Vorbis-encoded content made up only 12%.[
1]
In the commercial sector, Vorbis support is on the rise. Many
video game titles store game audio in the Vorbis format. Vorbis is supported on many consumer electronics like digital audio players as well. Many popular software players support Ogg Vorbis playback either natively or through an external
plugin. Another indication of Ogg Vorbis's increasing popularity is the number of websites using it, such as
Jamendo or
Mindawn, as well as several national radio stations such as
Radio France,
CBC Radio and
Virgin Radio providing additional Vorbis
streams.
Many other
lossy audio codecs exist. For many applications, Vorbis has clear advantages over other modern codecs in that it is patent-free and has open-source implementations, and therefore is free to use, implement, or modify as one sees fit, yet produces smaller files than most other codecs at equivalent quality.
Some listening tests have attempted to find the best quality lossy audio codecs at certain bitrates.
Some conclusions made by recent listening tests:
* At 128kb/s the
most recent large scale test shows a four way tie between aoTuV Vorbis,
LAME-encoded
MP3,
WMA Pro and
QuickTime AAC, with each codec essentially transparent (sounds identical to the original music file).
* At mid-low bitrates, latest private tests (
80 kb/s,
96 kb/s) shows that aoTuV Vorbis has a better quality than other lossy audio codecs (AAC (HE and LC), MP3,
MPC, WMA).
* At high bitrates (more than 128 kb/s), most people do not hear significant differences. However, trained listeners can often hear significant differences between codecs at identical bitrates, and aoTuV Vorbis performs very well, i.e. better to other formats such as AAC, MP3, and MPC (see this
180 kb/s test on classical music).
Many of these results, however, are difficult to keep up-to-date due to the ever-evolving nature of the codecs.
Listening tests
Listening tests are normally carried out as
ABX tests, i.e., the listener has to identify an unknown sample X as being A or B, with A (the original) and B (the encoded version) available for reference. The outcome of a test must be statistically significant. This setup ensures that the listener is not biased by his/her expectations, and that the outcome is not likely to be the result of chance. If sample X can be identified reliably, the listener can assign a score as a subjective judgement of the quality. Otherwise, the encoded version is considered to be
transparent. Below are links to several listening test results.
*
2005-2006 Public group test of Lame MP3, Vorbis AoTuV, iTunes AAC, Nero AAC, and WMA Pro at ~135 kbit/s nominal. Results suggest that further group testing at this bitrate is unnecessary because all codecs are statistically tied near transparency.
*
2005, July comparison - AAC vs MP3 vs Vorbis vs WMA at 80 kbit/s. States that aoTuV beta 4 is the best encoder for either classical or various music in this bitrate, and that its quality is comparable to the
LAME ABR MP3 at 128 kbit/s.
*
2005, August comparison - AAC vs MP3 vs Vorbis vs WMA at 96 kbit/s. States that aoTuV beta 4 and AAC are tied as the best encoders for classical music in this bitrate, while aoTuV beta 4 is the best encoder for pop music, even better than LAME at 128 kbit/s.
*
2005, August comparison - MPC vs VORBIS vs MP3 vs AAC at 180 kbit/s. An
audiophile listening test, which states that, for classical music, aoTuV beta 4 has 93% percent probability of being the best encoder, tied with MPC. MPC is tied with both Vorbis, in the first place, and LAME in the second.
*
2003-2004 comparisons of MP3, Vorbis, AAC, etc at a number of bitrates.Vorbis nominal bitrate at different quality levels (effective bitrate may be slighty different)
| Quality | Nominal Bitrate |
|---|
| Official Xiph.org Vorbis! aoTuV beta 3 and later |
|---|
| -q-2 | not available | 32 kb/s |
| -q-1 | 45 kb/s | 48 kb/s |
| -q0 | 64 kb/s | |
| -q1 | 80 kb/s | |
| -q2 | 96 kb/s | |
| -q3 | 112 kb/s | |
| -q4 | 128 kb/s | |
| -q5 | 160 kb/s | |
| -q6 | 192 kb/s | |
| -q7 | 224 kb/s | |
| -q8 | 256 kb/s | |
| -q9 | 320 kb/s | |
| -q10 | 500 kb/s | |
Given 44.1
kHz (standard
CD audio
sampling frequency) stereo input, the encoder will produce output from roughly 45 to 500 kbit/s (32 to 500 kbit/s for
aoTuV tunings) depending on the specified quality setting. Quality settings run from -1 to 10 (-2 to 10 for
aoTuV tunings) and are an arbitrary metric — files encoded at -q5, for example, should have the same quality of sound in all versions of the encoder, but newer versions should be able to achieve that quality with a lower bitrate. The bitrates mentioned above are only approximate; Vorbis is inherently
variable-bitrate (VBR), so bitrate may vary considerably from sample to sample.
Vorbis uses the
modified discrete cosine transform (MDCT) for converting sound data from the
time domain to the
frequency domain. The resulting frequency-domain data is broken into
noise floor and residue components, and then
quantized and
entropy coded using a
codebook-based
vector quantization algorithm. The decompression algorithm reverses these stages. The noise floor approach gives Vorbis its characteristic analog noise-like failure mode (when the bitrate is too low to encode the audio without perceptible loss), which many people find more pleasant than the metallic warbling in the MP3 format.
Many users feel that Vorbis reaches
transparency (sound quality that is indistinguishable from the original source recording) at a quality setting of -q5, approximately 160 kbit/s. For comparison, it is commonly felt that MP3 reaches transparency at around 192 kbit/s, resulting in larger file sizes for the same sound quality.
Various
tuned versions of the encoder (Garf, aoTuV or MegaMix) attempt to provide better sound at a specified quality setting, usually by dealing with certain problematic waveforms by temporarily increasing the bitrate. The most consistently cited problem with Vorbis is
pre-echo, a faint copy of a sharp attack that occurs just before the actual sound (the sound of
castanets is commonly cited as causing this effect). Most of the tuned versions of Vorbis attempt to fix this problem and to increase the sound quality of lower quality settings (-q-2 through -q4). Some tuning suggestions created by the
Vorbis user community (especially the
aoTuV tunings) have been incorporated into the 1.1.0 release.
The Vorbis format supports
bitrate peeling for reducing the bitrate of already encoded files, and an experimental implemention of this can be found
here. However, re-encoding files at a lower bitrate will preserve more quality than this experimental bitrate peeler.
Vorbis streams can be encapsulated in other media
container formats besides Ogg. A commonly used alternative is Matroska.
Metadata
Vorbis
metadata, called
comments, support metadata 'tags' similar to those implemented in the
ID3 standard for MP3. The metadata is stored in a vector of eight-bit-clean strings of arbitrary length and size. The size of the vector and the size of each string in bytes is limited to 2
32-1 (about 4.3
billion, or any integer that can be expressed in 32 bits). This vector is stored in the second header packet that begins a Vorbis bitstream.[
2]
The strings are assumed to be encoded as
UTF-8. Music tags are typically implemented as strings of the form "[TAG]=[VALUE]", for instance, "ARTIST=The John Smith Band". Since there is no strict field definition as in ID3, users and encoding software are free to use whichever tags are appropriate for the content. For example, an encoder could use localized tag labels, live music tracks might contain a "Venue=" tag or files could have multiple genre definitions. Most applications also support common de facto standards such as discnumber and
Replay Gain information.
Knowledge of Vorbis's specifications is in the
public domain. Concerning the specification itself, Xiph.org reserves the right to set the Vorbis specification and certify compliance. Its libraries are released under a
BSD-style license and its tools are released under the
GPL (GNU General Public License). The libraries were originally released under the
GNU Lesser General Public Licence, but a BSD licence was later chosen with the endorsement of
Richard Stallman [
3]. The
Xiph.org Foundation states that Vorbis, like all its developments, is completely free from the licensing or
patent issues raised by other
proprietary formats such as
MP3. Although Xiph.org says it has conducted a patent search that supports its claims, outside parties (notably engineers working on rival formats) have expressed doubt that Vorbis is free of patented technology [
4].
Xiph.org maintains that it was privately issued a legal opinion subject to
attorney/client privilege. It has not released an official statement on the patent status of Vorbis, pointing out that such a statement is technically impossible due to the number and scope of patents in existence and the questionable validity of many of them. Such issues cannot be resolved outside of a court of law. Some Vorbis proponents have derided the uncertainty concerning the patent status as "
FUD": misinformation spread by large companies with a vested interest.
Ogg Vorbis is supported by several large
digital audio player manufacturers such as
Samsung,
Rio,
Neuros,
Cowon and
iRiver. Many feel that the growing support for the Vorbis codec within the industry supports their interpretation of its patent status, as multinational corporations are unlikely to distribute software with questionable legal status. The same could be said about its growing popularity in other commercial enterprises like mainstream computer games.
Since the Vorbis libraries are available under a
BSD licence (a
free software approved licence) and the format itself is
accepted as not covered by patents, several video game developers[
5] have chosen to use Vorbis in their games rather than pay for patent-encumbered competitors like MP3. Some PC examples include:
*
id Software's
Doom 3*
Lionhead's
The Movies*
Epic Games'
Unreal Tournament 2003 and
Unreal Tournament 2004*
Crytek's
Far Cry*
Croteam's
Serious Sam and
Serious Sam II*The PC port of
Microsoft's
Halo*
GTA San Andreas*
Myst IV: Revelation*
Schizm*
Operation Flashpoint*
Duke Nukem: Manhattan Project*
America's Army*
Trackmania*The PC port of
Mega Man X8*
Ys VI: The Ark of Napishtim*
SEGA's
Phantasy Star Online Blue Burst*
CrimsonlandA full list can be found at
Xiph.org WikiIn April 2006,
RAD Game Tools, one of the leading Game Development Toolkit makers, announced that their Miles Sound System toolkit fully supports Vorbis.
In addition to the PC, Vorbis has also gained popularity on video game consoles due to the need for developers to compress game data. Known console games which use Vorbis include:
*
Free Radical Design's
TimeSplitters: Future Perfect (GameCube version only)
*
Capcom's
Mega Man X Collection (GameCube version only)
Hardware
Tremor, a version of the Vorbis decoder which uses
fixed-point arithmetic (rather than
floating point), was made available to the public on
September 2,
2002 (also under a
BSD-style license). Tremor, or platform specific versions based on it, is more suited to implementation on the limited facilities available in commercial portable players. A number of versions that make adjustments for specific platforms and include customized optimizations for given embedded microprocessors have been produced. Several hardware manufacturers have expressed an intention to produce Vorbis-compliant devices, and new Vorbis devices seem to be appearing at a steady rate, especially in
South Korea, although availability may differ from country to country.
The
VorbisHardware node at the xiph.org wiki has an up-to-date list of Vorbis-supporting hardware, such as portables, PDAs, and microchips. Most digital audio players supported by
Rockbox, an open-source firmware project, are capable of decoding Vorbis.
Software
Software supporting Vorbis exists for many platforms. Although
Apple iTunes does not natively support Vorbis, Xiph.org
provides a QuickTime component which can be used in iTunes and
QuickTime on both
Microsoft Windows and
Mac OS. On Windows,
DirectShow filters exist to decode Vorbis in multimedia players like
Windows Media Player and others which support
DirectShow. Vorbis is well-supported on the
Linux platform in programs like
XMMS,
xine, and many more.
More information about Vorbis-supporting software can be found at the
VorbisSoftwarePlayers node at the xiph.org wiki. Users can test these programs using the list of Vorbis audio streams available at
the Vorbis streams page on the same wiki.
*"Ogg" is not named after the witch
Nanny Ogg in
Terry Pratchett's
Discworld books; see
Ogg for the correct definition. However, "Vorbis"
is named after another
Discworld character, High Priest Vorbis in
Small Gods.
*Sound samples on
Wikipedia and many other
Wikis use the Ogg Vorbis format for sound media.
*
Help with Ogg Vorbis and playing sound files on Wikipedia
*
Ogg bitstream format
*
Vorbis comment, metadata format used by Vorbis
*
Speex,
speech codec*
FLAC,
lossless audio codec*
Theora,
video codec*
Tarkin, video codec
*
XSPF,
playlist format
*
Comparison of audio codecs*
Xiph QuickTime Components, official QuickTime implementation
*
FreeCast, peer-to-peer Vorbis streaming
*
Xiph.org Foundation*
Ogg Vorbis site - Xiph reference implementation.
**
Listening test comparing Vorbis to MP3, RealAudio, WMA, etc.**
Ogg Vorbis wiki - mostly developer information**
List of Vorbis audio streams**
Vorbis FAQ**
Vorbis documentation*
aoTuV - aoTuV Vorbis tuned implementation.
*
modified discrete cosine transform Description*
Christopher "Monty" Montgomery (main developer) interview, slashdot.org*
Ogg Vorbis binaries at Rarewares*
Ogg Vorbis Hydrogenaudio wiki *
jamendo.com : Largest archive of Creative Commons licensed Ogg Vorbis music*
Vorbis DirectShow filter for Windows*
QuickTime component for QuickTime and iTunes*
Ogg Vorbis acceleration project - Optimized versions of aoTuV Vorbis encoder and other SSE/MultiThread optmizations.
