I seldom write reviews for books on Amazon, but I just had to do it for this excellent book. This book is by far the best resource for building enterprise caliber voice applications. I have purchased 5 other VXML books and none of them delivers the information that this book does. This book not only delivered the content I needed, but it's comprehensive and deals with complex issues in a simplified fashion. The real highlights of this book are Chapters 4 and 5. Chapter 4 explains Enterprise applications, possible architectures for dealing with state, and interacting with a backend. Chapter 5 deals with more of a deployment and physical architectural angle. Get it for those two chapters alone.
The only thing that I found missing was techniques for versioning a voice application (No not the CVS source-code control type versioning). I mean having the same backend server provide different content to the same voice browser and techniques on structuring your application to smoothly transition from one set of content to another. For example, the voice browser cache is full of old static content - how can I gracefully switch to new content without forcing a dump of the existing cache through the HTTP headers or some other external mechanism - It has to do with using relative pathing beneath the application root document for vxml scripts and static content. Some information about Browser to Browser interactions might also be nice - but I recognize the VBI specs are just emerging.
Anyway, anyone thinking of building a real voice application with speech recognition and integration to backend applications and data should definitely add this book to their library. It has helped me tremendously. Two thumbs up!