This book could also be titled Secrets of an audio programming guru. Tashev provides valuable practical insights by sharing his personal experiences with building, testing, and tuning advanced audio processing algorithms.

There are times when an audio developer needs to quickly implement and test a specific audio processing task, without necessarily having deep knowledge of the theory of operation. Blind use of suitable software libraries--if any exist--rarely leads to the desired results; also, some theoretical background is necessary, especially for the tuning part of the implementation. Although there are numerous books and papers on the topics covered in this book, 99 percent of them require weeks--if not months--of study, in order to extract and distill the necessary information for the practical use and tuning of a particular audio processing algorithm. Tashev’s book reduces the study time to days or even hours, without sacrificing clarity and rigor. Furthermore, the publisher’s Web site provides several algorithm implementations in MATLAB (http://www.wiley.com//legacy/wileychi/tashev_sound/).

The book consists of eight chapters. Chapter 1 explains the motivation for the presentation that follows, defines the target audience, and presents the structure of the material. Any book that discusses complex topics has some prerequisites, and this book is no exception. In chapter 2, the author introduces some of these, with a certain bias toward frequency domain processing, which is at the core of most modern audio processing algorithms. As an example of practicality, readers will get the same practical results from chapter 7 in Crochiere and Rabiner’s book on multirate filtering [1] and page 39 in Tashev’s work. (This is not to say that chapter 7 is unnecessary, but that Tashev concisely presents what is essential for fast and accurate practical implementation in one page.)

Chapter 3 includes a short primer on acoustics and a detailed overview of microphones, especially their construction and manufacturing tolerances--important issues in the context of arrays. Chapter 4 introduces single-channel noise reduction techniques within a statistical signal processing framework. The chapter includes sections on signal estimation, noise suppression, and suppression rules, including the Wiener filter, the more elaborate Ephraim-Malah scheme, and probability-based rules. Other sections cover voice activity detectors, the general architecture of a noise suppressor, and specialized noise suppression systems, such as adaptive noise cancellation. The last section provides practical tips for tuning and implementing noise suppressors.

Chapter 5 presents techniques for designing both fixed and adaptive microphone arrays. While many of the theoretical derivations can be found elsewhere, their practical significance is rarely presented as clearly as it is in this chapter. Moreover, there is a section that deals with how to handle microphone tolerances and how to implement self-calibration systems--the level of detail presented here is unique to this book. In chapter 6, the focus is on the application of array techniques to sound localization and tracking. Most major algorithms are covered, including the multiple signal classification (MUSIC) algorithm, the phase transform (PHAT), and Bayesian techniques. The chapter ends with sections on post-processing algorithms and invaluable practical tips.

Chapter 7 deals with acoustic echo cancellation (AEC). While very thorough and practical, it does not provide design recipes for AEC. However, Tashev gives enough insight into alignment, handling nonlinearities, voice activity detection importance, and echo suppression that a practitioner can use this section to improve an existing AEC. The book ends with a very interesting chapter on de-reverberation. Tashev considers deconvolution and multi-microphone sound capturing--that is, beamforming. It provides practical advice on how to proceed and what to expect from the various algorithms.

This book is unique in the way it teaches practical details of several advanced audio processing procedures. It is not only appropriate for readers involved in digital audio, but also for those who consider digital audio the next great technical challenge.