Perceptual Audio Hashing Functions

  • PDF / 532,733 Bytes
  • 14 Pages / 600 x 792 pts Page_size
  • 30 Downloads / 245 Views

DOWNLOAD

REPORT


erceptual Audio Hashing Functions ¨ Hamza Ozer Department of Electrical and Electronics Engineering, Bo˘gazic¸i University, 34342 Bebek, Istanbul, Turkey National Research Institute of Electronics and Cryptology, Tubitak, 41470 Gebze, Kocaeli, Turkey Email: [email protected]

¨ Bulent Sankur Department of Electrical and Electronics Engineering, Bo˘gazic¸i University, 34342 Bebek, Istanbul, Turkey Email: [email protected]

Nasir Memon Department of Computer and Information Science, Polytechnic University, Brooklyn, NY 11201, USA Email: [email protected]

Emin Anarım Department of Electrical and Electronics Engineering, Bo˘gazic¸i University, 34342 Bebek, Istanbul, Turkey Email: [email protected] Received 13 September 2004; Revised 16 February 2005; Recommended for Publication by Mark Kahrs Perceptual hash functions provide a tool for fast and reliable identification of content. We present new audio hash functions based on summarization of the time-frequency spectral characteristics of an audio document. The proposed hash functions are based on the periodicity series of the fundamental frequency and on singular-value description of the cepstral frequencies. They are found, on one hand, to perform very satisfactorily in identification and verification tests, and on the other hand, to be very resilient to a large variety of attacks. Moreover, we address the issue of security of hashes and propose a keying technique, and thereby a key-dependent hash function. Keywords and phrases: perceptual audio hashing, content identification, singular value decomposition, least-square periodicity estimation.

1.

INTRODUCTION

In this study, we develop algorithms for summarizing a long audio signal into a concise signature sequence, which can then be used to identify the original record. We call this signature the perceptual hash function, because it is purported to reflect the perceptible component of the content. In other words, we aim to obtain audio hash functions that are insensitive to “reasonable” signal processing and editing operations, such as filtering, compression, sampling rate conversion and so forth, but that are otherwise sensitive to the change in content. Such perceptual hash functions can be used as a tool to search for a specific record in a database, to verify the content authenticity of the record, to monitor broadcasts, to automatically index multimedia libraries, to This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

detect content tampering attacks, and so forth [1]. For example, in database searching and broadcast monitoring, instead of comparing the whole sample set, the hash sequence would suffice to identify the content in a rapid manner. In tamper proofing and data content authentication applications, the hash values of the applicant object are compared with hash values of the stored ones. In the watermarking context, it is desirable to e