Audio watermark to stop data pirates

Designing Smarter Technologies

A collaborative research project between Deakin and industry aims to provide a new weapon in the fight against the illegal use of copyrighted data.

The news that season seven of the HBO hit TV show ‘Game of Thrones’ was pirated more than one billion times within seven weeks of its release in 2017 made headlines, while illegal sharing of music is a worldwide problem. In Australia alone, around 2.8 million people download music illegally via file sharing networks and one billion songs are illegally traded by Australians every year.

Demand for new ways to trace unlawful downloading and sharing of multimedia data has led to a team of researchers from Deakin University’s School of Information Technology and industry partner Flag Explore Pty Ltd receiving an ARC Linkage Project Grant to advance next-generation audio watermarking techniques that can trace illegal copying and distribution of data containing a sound component, such as music and video.

Deakin’s Professor Yong Xiang, Associate Head of School (Research), School of Information Technology, who led the successful bid, will work with former Deakin Professor Wanlei Zhou (now of University of Technology Sydney) and Deakin’s Professor Gleb Beliakov and Dr Longxiang Gao in partnership with Flag Explore to develop inaudible, robust and high-capacity audio watermarking technology.

Professor Xiang said the project’s outcomes would provide a ready-to-use audio watermarking solution for real-world applications and help prevent financial and job losses in the Australian multimedia industry.

“Multimedia piracy is a serious problem and the financial loss caused by illegal multimedia data downloading and sharing is enormous,” Professor Xiang said.

“It not only harms the intellectual property owners, but causes significant damage to the Australian economy due to revenue and job losses. This issue is particularly important to our industry partner, Flag Explore, which provides online live streaming services to end-users. Its business largely relies on the effective protection of its copyrighted audio and sound video data.”
Professor Yong Xiang
Professor Yong Xiang Associate Head of School (Research) School of Information Technology

Digital audio watermarking hides watermark data such as publisher information, user identity and file transaction/downloading records into the audio signal without affecting its normal usage. Using a secret key, the owner of the data, or law enforcement agencies, can extract the watermark data to trace the source of illegal distribution.

Professor Xiang said an effective and practical watermarking scheme should include the important characteristics of imperceptibility, robustness against conventional attacks and high embedding capacity.

“Although a number of methods have been proposed for audio watermarking, built upon various mechanisms, none of them can ensure imperceptibility, robustness, and high embedding capacity simultaneously, and they are particularly vulnerable to collusion attacks. As a result, their application in practice is restrictive,” he explained.

“There is an urgent need to develop inaudible, robust and high-capacity audio watermarking techniques that will deter multimedia data infringement and help authorities trace and punish pirates.

“The audio watermarking technology developed in this project will provide a ready-to-use solution to tackling copyright infringement and build a solid foundation for developing other multimedia piracy tracing products.”

 

Published by Deakin Research on 5 June 2018