Digital audio watermarking

Given the vastness of cyberspace and the daily increase in the number of sites that use sound and video, it has become apparent to the music industry that securing legal protection for sound recordings is only one small step towards protecting music on the Internet . Enforcing those rights is a different matter entirely, especially when dealing with music which is little more than computer data stored on a computer disk.

In June 1997, the International Federation of the Phonographic Industry (IFPI ) issued a Request for Proposals for an "Embedded Signalling System" which would be used to manage and track the use of digital music in the European music market . If successful, this watermarking technology would later be introduced into the US entertainment market, because, according to Paul Jessop (director of Technology at IFPI)

 

"It will be a decidedly crucial factor towards the entertainment industry's willingness to advance DVD Video beyond its current existence as a technophiles" medium, to introduce DVD Audio to all, and to support the Internet as a retail outlet for CD -quality music" [40].

This project became known as the MUSE project, and eight to twelve digital audio technology companies answered IFPI 's request for proposals before the deadline at the end of June, all of whom have so far remained anonymous. Companies who are currently developing watermarking software include the US-based companies a2b music - a subsidiary of AT&T Laboratories, Liquid Audio and Intersect, Europe-based AudioSoft, and London-based Cerberus.

What is digital audio watermarking?

Research into digital technology, undertaken so far by various companies around the world, has developed various means of watermarking data in digital formats, such as on CDs, MiniDisc, DAT and sampled sound files . The idea is to embed a code within the audio signal (or on the image of a video on CD -ROM) which can be recovered using the appropriate technology, without affecting the quality of the product. Only a small amount of the total data of the track is allocated to the watermark - if one second of stereo digital audio is represented by 1,411,200 bits, the maximum throughput of most commercially-available watermarking processes amounts to only 0.001417% of the total (assuming that the watermark is present throughout the track).

Since the watermark is an integral part of the recording, it cannot be removed without seriously damaging the sound quality. Since the digital watermark is inaudible, any information can be contained therein, including names of the song, artist, writers, album, record company etc. The code (or datastream) embedded in the audio could also contain information regarding the e-mail address and identity of the person who downloaded the track and the serial number of the audio recorder used in the transaction.

There are several applications of this technology which are of benefit to the music industry

Royalty collections/broadcasting monitoring

Watermarking will automate the process of tracking performance royalties through the monitoring of broadcasting . In theory, collection agencies such as BMI, ASCAP, MCPS or SESAC could have offices linked via computer to radio and TV stations. When a song is played through the decoding key on a computer, an entry would be registered on a database, allowing the station to compile an accurate report of songs broadcast which can be forwarded to the collection agency without the need to manually complete logs. This method also guarantees a more accurate payment of royalties and cuts out the amount of red tape currently involved in broadcast monitoring.

Repertoire stock control/sales tracking

Since thousands of recordings are released every year, the use of digital audio watermarking would greatly increase the efficient tracking of stock through database systems. Large retail outlets would benefit from sales data by

• allowing a constant supply of best-selling titles without danger of losing revenue through being sold out of the most popular titles
  
  
• keeping an accurate record of sales from one week to the next, assisting in decisions about what titles to re-order, which titles/artists are likely to sell in the future, which stock lines to adapt or cancel, and which titles to enter into special promotions.
  
  
• compiling demographic and other market information which will be of benefit to both retail outlets and record/distribution companies - i.e. who is buying what, when and where? This data can help companies target specific audiences, increasing the efficiency and revenue of the music industry.

Authentication

Watermarking clearly labels the copyright owner deep within the audio signal and in the event of a standard being developed for consumer hardware to read basic information embedded in the data, watermarking can assure the consumer of a product's authenticity. A bootleg, for example would indicate its origin in a promotional product or indicate its bootleg status by a lack of information.

As far as the Internet is concerned, companies have been developing special programs which search the World Wide Web for sites that include audio or video clips, and verify the intellectual copyright of those clips by reading a watermark . Intersect Inc. Is a US-based company which has developed MusicTrak, a monitoring and reporting system which tracks the use of music on the Internet. Using Audio Video Scan (AVS) technology, Intersect can search up to eleven million Web pages per day for MPEG, RealAudio, Liquid Audio and other media formats used to distribute music and video. AVS technology also supports new media file formats as they become available and can search the entire Internet (currently around 60 million sites) in just two weeks. [41]

Intersect's MusicTrak system comprises of three elements

1. WebCrawling subsystem
2. Indexing Subsystem
3. Search Engine Subsystem

The first of these subsystems, the WebCrawling Subsystem also known as a spider or bot, is an intelligent software agent that runs without an operator, searching the Internet and retrieving data by

1. crawling the Web, sending and receiving replies to and from Web Servers
   
   
2. retrieving documents, obtaining a copy of every page on the Web for the purpose of scanning for music files
   
   
3. searching for new URLs and other references embedded within Web pages
   
   
4. delivering documents in the form of HTML pages to the Indexing subsystems, which register the Web address of every form of media clip available on the Internet . This database can then be searched on behalf of record companies or collection agencies to discover who is using their music, where it can be found, what use is being made of it and whether it is a legitimate or illegal copy. Legal action can then be taken on Internet pirates.

A system diagram of the Intersect MusicTrak monitoring solution [42]