MetadataI recently had to do some research on metadata in order to see how to control the distribution of mp3 and other lossy formats. Some of the information is hard to wrap your head around, but I think I have simplified it to the point of anyone being to understand most of it. My entire bulletin I wrote is after the jump. If you have any comments on things I messed up or to give more insight, just hit the comment button and type away!

Metadata: The Format Wars Continue

By: Dave Towne 3/29/2007

· Metadata

o Metadata is described as “a type of data where something being described is data”

§ More simply put: information describing data

o We use metadata to help describe certain data that may take up too much space or as a code to reference something.

§ We use it everyday

· An example would be a postal zip code

· 44115 is metadata for an area of Cuyahoga County in Cleveland

o There are plenty of formats when it comes to metadata and how it is stored/read

· Audio Formats and their metadata

o Audio CD

§ CDDB

· CDDB is a database for software to lookup CD information such as:

o Track artist

o Track name

o Album name

o Album Year

· CDs store information on when tracks start, how long they last, and how long the album lasts, known as metadata. This information is referenced with an online database called the CDDB and all the new metadata is downloaded to a program such as iTunes, Windows Media Player, and Winamp.

§ CD-Text

· This is actually burned onto either the lead-in area of an audio CD or in its sub-channels R through W.

· The CD-Text can actually be used to help create CDDB information when a CD with CD-Text is ripped into a program like iTunes.

§ ISRC (and other PQ data)

· ISRC stands for International Standard Recording Code

· The ISRC is simply an international standard for uniquely identifying sound and music video recordings.

· Consists of 12 characters in the form of CC-XXX-YY-NNNNN

o CC = Country code

o XXX = Registrant code

o YY = Year of registration

o NNNNN = A unique 5-digit number for a song

· These are added to the sub-channel Q on Audio CDs.

o MP3 & AIFF

§ MP3s do not directly have sub-channels for data because they are one continuous file and not a collection within a medium (tracks on a CD). Therefore they cannot keep PQ data, CD-Text and CDDB.

§ They use a tag, embedded into the file, to reference the metadata.

§ Tags describe various things about the file such as:

· Artist name

· Track name

· Album year

· Lyrics

· Composer

§ The current standard for MP3 tags is the ID3 v2.3

· ID3 is often referred to as a “container format” because it can evolve with the change of the file

o This means data can be added or removed by any time by any user.

§ Although they contain such information, ID3 tags only stay with the file they are attached to. They are not linked to a database

· If you give your buddy an MP3 and he changed the year content in the ID3 tag, your MP3 will now show the change


o AAC

§ A lossy format for digital music that is known for small file sizes with better sound than the MP3 format

§ Uses a variable bit rate for encoding

§ Restricts files by using a Fairplay, “DRM” system by using a .mp4 container attached inside the .AAC file.

§ AAC is much like the MP3 and sometimes uses the ID3 v2.3 tag

§ Although, companies like iTunes have been said to use their own tagging scheme called “atoms”

· Atoms look basically same as the ID3 to the user, but they are not cross compatible and do not cross reference when something is ripped or re-encoded.

· Watermarking Audio Files

o As stated above, many of the most popular digital formats used by professionals and consumers do not have un-editable information like that which is stored on a CD. (PQ, ISRC and UPC codes)

o Watermarking is a fairly new concept for audio and basically allows the creator to track if their digital media is being distributed illegally.

o A simple way to look at watermark encoding files is to relate it to dithering.

§ Just like dithering, watermark encoding injects a noise into the audio content

§ This noise includes digital data, un-editable, known as a key.

§ The key is spread during times that the encoder sees as optimal masking points in the music so that the noise is not audible and so that the noise does not disappear when a user re-encodes or re-samples the audio content.

§ If someone actually finds the digital key code and tampers with it, the audio file will be distorted heavily because the code is attached to the actual audio.


References

Metadata

http://en.wikipedia.org/wiki/Metadata

http://www.tvtechnology.com/features/audio_notes/f-tc-metadata.shtml

http://www.dolby.com/assets/pdf/tech_library/17_.AllMetadata.pdf

CDDB

http://en.wikipedia.org/wiki/CDDB

http://www.gracenote.com/

CD-Text

http://en.wikipedia.org/wiki/CD_Text

http://web.ncf.ca/aa571/cdtext.htm

http://www.feurio.com/English/faq/faq_vocable_cdtext.shtml

ISRC

http://en.wikipedia.org/wiki/International_Standard_Recording_Code

http://www.ifpi.org/content/section_resources/isrc.html

http://www.riaa.com/issues/audio/isrc_faq.asp

MP3

http://en.wikipedia.org/wiki/MP3

http://www.howstuffworks.com/mp3.htm

AIFF

http://en.wikipedia.org/wiki/AIFF

http://www.borg.com/~jglatt/tech/aiff.htm

http://www.digitalpreservation.gov/formats/fdd/fdd000005.shtml

AAC

http://en.wikipedia.org/wiki/Advanced_Audio_Coding

http://www.apple.com/quicktime/technologies/aac/

http://www.audiocoding.com/modules/wiki/?id=AAC

Watermarking

http://www.ini-graphics.net/press/topics/1999/issue1/1_99a03.pdf

http://www.musictrace.de/products/contentmark.en.htm

http://activatedcontent.com/_ActivatedAudio/Technologies/default.aspx

http://www.igd.fraunhofer.de/igd-a8/syscop/audio.html

http://www.bluespike.com/faqongiovanni.html