A Voice-Based Biometric Watermarking Scheme For Digital Rights Management of 3D Mesh Models
AuthorMotwani, Rakhi Chetandas
StatisticsView Usage Statistics
Widespread use of 3D artwork has necessitated the need for managing digital rights of content owners due to illegal peer-to-peer (P2P) distribution of artwork, which has been one of the major sources of revenue loss for the art industry. Existing Digital Rights Management (DRM) systems attempt to provide an anti-piracy framework that restricts the use of content to its rightful user. However, limitations of technology have consequently led to solution designs that are either very restrictive (i.e. device-limiting or usage restrictions) or only succeed in discouraging unlawful distribution by employing tracing mechanisms. The objective of this dissertation is to contribute towards DRM implementations to ensure fair rights management such that the needs of both the authors and consumers are balanced. The proposed biometric-based watermarking scheme allows DRM systems to enforce copyright protection by imposing individual-limiting usage rights, thereby eliminating any device dependency or usage restrictions. The proposed technique embeds the identity of the consumer in the form of a voice print, into the graphic content. This voice print serves as a watermark and is created by using a statistical model (Gaussian Mixture Model) representation of the consumer's voice sample. Mel-frequency cepstral coefficients, representing feature vectors of the speaker from the speech signal, are used to generate the statistical model. Comparison of the embedded watermark with feature vectors extracted from a newly acquired voice sample from the consumer enable the biometric-based DRM system to verify the authenticity of the consumer for legitimate access and usage of the 3D artwork. The scope of this dissertation is limited to 3D mesh models.