Browse Books

Reviewer: Soubhik Chakraborty

Audio content analysis (ACA) is actually a subtopic of the broader music information retrieval (MIR) research area. This subtopic deals with bringing out musical and perceptual properties directly from the audio signals to improve human-computer interaction (HCI) with digital audio signals. A good understanding of ACA assists in the design of intelligent MIR applications and content-adaptive audio processing systems. In the author's own words, "ACA is a multidisciplinary research field" requiring knowledge from "different research fields such as musicology and music theory, (music) psychology, psychoacoustics, audio engineering, library science, and last but not least computer science for pattern recognition and machine learning." Chapter 1 introduces ACA and chapter 2 covers the fundamentals of audio signals and signal processing. The major topics covered in the remaining chapters of the book include instantaneous features (such as statistical properties, spectral shape, and signal properties), intensity, tonal analysis, temporal analysis, alignment, musical genre, similarity and mood, audio fingerprinting, and music performance analysis. The author provides a very handy appendix on convolution properties, Fourier transforms, principal component analysis, and software for audio analysis. The book includes many salient features: It is a very good guide to ACA and its application in signal processing and music informatics. It treats various characteristics of musical information separately, including pitch, harmony, tempo, key, tonality, and timbre. It includes a helpful review of the basics of audio signal processing, music theory, and psychoacoustics (making it useful as an introductory text). It analyzes and compares different algorithms for the same task. Its companion website (http://www.audiocontentanalysis.org/) includes invaluable MATLAB programs that are freely downloadable. It concludes with a comprehensive bibliography. The author is an acknowledged expert in the music industry. This book will not only greatly help undergraduate and graduate ACA students, but will also be a boon to music researchers and music industry experts alike. The book is simply a treasure for music analysts, and I would strongly recommend it for any scientific library. It does not, however, focus on speech signals; as such, automatic speech recognition, although within the scope of ACA, has been omitted. To use the book profitably, an elementary knowledge of digital signal processing (DSP) is necessary. More reviews about this item: Amazon Online Computing Reviews Service

Reviewer: Vladimir Botchev

The major positive traits of this tiny book are that it gathers in one place the information that up to now had been scattered in papers, open-source code descriptions, and specialized Internet forums (mostly academic ones), and that it provides an example of MATLAB code on the book's Web site (http://www.audiocontentanalysis.org/) that is easy to understand and use. After an introductory chapter, the book begins with a chapter (2) devoted to elementary concepts in digital audio signals and their basic transforms (Fourier, constant Q , and auditory filter banks). Chapter 3 introduces the so-called instantaneous feature, in other words, a numeric qualifier for a short segment of the signal being analyzed. The emphasis is on statistical qualifiers, such as moments, and spectral shape qualifiers. The descriptions are concise and ready to use in applications that would need these features. Chapter 4 gives a short description of some intensity features, such as signal envelopes. Tonal analysis, which includes pitch processing, is given more thorough treatment in the next chapter. Chapter 6 presents details in the area of temporal analysis, such as tempo, beats, and onset detection. Chapter 7 concludes the discussions on features and their extraction with a discussion of algorithms for time alignment, including dynamic time warping. The last three chapters are devoted to basic applications, some of which are well known to smartphone users, such as musical genre recognition and music similarity, described in chapter 8. Chapter 9 gives a glimpse into audio fingerprinting. The last chapter introduces music performance analysis. There are four appendices, a short description of convolution properties, a lengthy description of Fourier transforms, a two-page courtesy mention of principal component analysis, and a quite useful summary of some of the major (and best so far) open-source software platforms that are either devoted to or usable for audio content analysis. Overall, this is a very practical book. It's a good source of concise information on many topics in audio analysis, and I recommend it for practitioners of digital audio. Online Computing Reviews Service