For anyone involved in microphone processing, acoustic echo cancellation, acoustic direction finding, and many other similar topics, this is the book. The same holds for anyone who is simply interested in those topics, but these readers should be warned that this book is not light reading, and its price is rather high.

The book is concerned with topics on which it is hard to find detailed information on paper, and sometimes in any format. It is a fine collection of research work, and if some topics seem to lack enough detail, it should be understood that many of the detailed technical solutions in these areas are either patented or have patents pending. The book contains 15 chapters, organized in six parts. The first chapter is a basic introduction, whose contents are related to Parts 1 and 2. It introduces the acoustic echo problem and basic adaptive filtering–based solutions for it.

Part 1 is concerned with single-channel acoustic echo cancellation (AEC). First, chapter 2 details the derivation of the fast affine projection algorithm for adaptive filtering. Issues such as numerical stability and implementation complexity are discussed. Chapter 3 deals with adaptive filtering in subbands, the approach used in many acoustic echo cancellers used in the industry. In this chapter, important issues related to fixed-point implementation are discussed. Chapter 4 covers a frequency domain adaptive filtering solution to the AEC problem. While it has some similarities to the previously discussed filterbank approach, block processing frequency domain adaptive filtering also has some advantages, for example, in terms of signal delay. Chapter 5 deals with double-talk detection. This is a crucial topic in acoustic echo cancellation, since not taking it into account may make an acoustic echo canceller totally unusable. Some of the results in this chapter were published in the IEEE Transactions on Speech and Audio.

Part 2 is concerned with multichannel (stereo) acoustic echo cancellation. This is a very challenging domain, since, in general, the problem does not have a unique solution. Chapter 6 introduces multichannel sound communication and multichannel time domain adaptive filtering. The theme of chapter 7 is frequency domain adaptive filtering for multichannel AEC. Chapter 8 details the implementation of a real-time stereophonic AEC, but without presenting actual code.

Part 3 is a single chapter about noise reduction in single-microphone systems, which have no noise sensor to help reduce unwanted disturbances. Chapter 9 discusses popular methods such as spectral noise subtraction, and gives some details that can help in the implementation of these methods.

Part 4 details the hot topic of microphone arrays--hot because, by directional means, the quality of the acquired sound can be substantially improved, and this method can be used in a wide range of applications, ranging from speech recognition to teleconferencing and music pickup to cellular phone sound pickup. This is another area in which most of the information is proprietary. This part begins with chapter 10, on superdirectional microphones, a superbly written and extremely detailed chapter. To the best of my knowledge, this is the first time such detailed information about superdirectivity and superdirective arrays has been published. Microphone arrays based on the principles discussed in this chapter can be found in some commercially available teleconferencing systems. Chapter 11 discusses problems and provides high-level solutions to the problem of steering a video camera with a microphone array. Similar systems are in production. The last chapter in this part, chapter 12, discusses nonlinear, model-based microphone arrays for speech enhancement. Besides the many teleconferencing applications that may be envisaged, a candidate application is speech pickup for speech recognition systems in a desktop environment.

Virtual sound is the topic of Part 5. Chapter 13 begins with an introduction to 3D audio and virtual acoustical environments. Head related transfer functions (HRTF), HRTF modeling, and algorithm implementation issues are reviewed. Chapter 14 is concerned with the creation of virtual sound sources by a small number of loudspeakers and the related problem of acoustic crosstalk.

Finally, Part 6 consists of chapter 15, which provides theoretical details and simulation results for blind source separation of speech signals. The trick is to recover independent sources given sensor (microphone) outputs in which the sources have been mixed by an unknown channel.

No other book treats these extremely interesting and important teleconferencing topics. The detailed contents and originality of the problems and solutions covered make this book a must for digital audio professionals involved in telecommunications systems design.