A Review of Audio Fingerprinting

Abstract

Abstract. An audio fingerprint is a compact content-based signature that summarizes an audio recording. Audio Fingerprinting technologies have attracted attention since they allow the identification of audio independently of its format and without the need of meta-data or watermark embedding. Other uses of fingerprinting include: integrity verification, watermark support and content-based audio retrieval. The different approaches to fingerprinting have been described with different rationales and terminology: Pattern matching, Multimedia (Music) Information Retrieval or Cryptography (Robust Hashing). In this paper, we review different techniques describing its functional blocks as parts of a common, unified framework.

A comparison of extended fingerprint hashing and locality sensitive hashing for binary audio fingerprints

“…If we require a successful search rate of 90%, this system will not work for bit errors greater than about 0.136. This corresponds with empirical observations that this method does not work well for higher bit errors [9,5]. In general, Equation 1 allows us to predict the performance of a fingerprint hash for any bit error rate.…”

“…This hash table is analogous to an inverted index [19,15] used in text retrieval. As such, this hash-based method is sometimes referred to as an inverted file index [5]. The data structure is illustrated in Figure 1.…”

“…We assume that a good fingerprint generation scheme results in an approximately uniform distribution of fingerprints, and that the correct answer to a query is likely within a certain range which can be determined empirically. (A review of fingerprint generation methods can be found in [5]. )…”

Audio Cover Song Identification and Similarity: Background, Approaches, Evaluation, and Beyond

“…Second Hand Songs 3 , Coverinfo 4 , Coverville 5 , Midomi 6 , Fancovers 7 , or YouTube 8 . Frequently, these sites also allow users to share/present their own (sometimes homemade) cover songs, exchange opinions, discover new music, make friends, learn about music by comparing versions, etc.…”

“…Therefore, genre classification is considered to have a low match specificity [10]. On the other hand, audio fingerprinting [7] is an example of a task with a highly specific match. This essentially consists in identifying a particular performance of a concrete song (exact duplicate detection).…”