Larger than an exhaustive IEEE tutorial paper, but smaller than an actual book, this “brief” focuses on frequency domain processing for speech enhancement. The style is that of a theorem (or rather statement) proof. Its conciseness and attention to the details in most of the derivations or proofs make it well suited for a crash course in frequency domain speech enhancement, with an exclusive bias toward theoretical developments.

Unfortunately, this theoretical formality is pushed to the limit. For example, in chapter 4 on multichannel speech enhancement including microphone arrays, there is not a single graph. Perhaps the reader is supposed to know the subject well enough, so as not to need graphs and diagrams, but why then would a reader well versed in the subject need to read that chapter at all? In fact, the book contains a total of only five figures, all in the introductory chapter. This is rather strange for a signal processing book, especially when one considers that array processing owes many of its unique features to specific geometries. In other words, this book falls short on the practical side.

If ones writes a thesis, and needs to fill it with impressive derivations, then this is the right book. If one needs to get an intuitive feel for how enhancement in the frequency domain works or hands-on experience by simulating algorithms, it would be better to look for Loizou’s book on speech enhancement [1]. That book has the same mathematical elegance as the current one, with the added bonus of implementation details and simulation results that are reproducible. Loizou’s book lacks a multichannel section, but given the fact that the current book’s multichannel section is overly formal, it is hard to look at this omission as a disadvantage when comparing the two.