In this paper a method consisting in embedding low-level music feature vectors as watermarks into a musical signal is proposed. First, a review of some recent watermarking techniques and the main goals of development of digital watermarking research are provided. Then, a short overview of parameterization employed in the area of Music Information Retrieval is given. A methodology of non-blind watermarking applied to music-content description is presented. The system architecture for the embedding and recovery of the watermarks, along with the algorithms implemented, are described. The robustness of the watermark implemented is tested against audio file processing, such as re-sampling, filtration, time warping, cropping and lossy compression. Procedures for simulating musical signal alteration are explained with a focus on the influence of lossy compression on the degradation of the embedded watermark. The advantages and disadvantages of the proposed approach are discussed. An outline of future applications of the methodology introduced is also included.
Authors
Additional information
- DOI
- Digital Object Identifier link open in new tab 10.1007/978-3-642-35647-6_27
- Category
- Publikacja monograficzna
- Type
- rozdział, artykuł w książce - dziele zbiorowym /podręczniku w języku o zasięgu międzynarodowym
- Language
- angielski
- Publication year
- 2013
