This book is a comprehensive and excellent introduction to the ever-expanding
field of Automatic Speech Recognition. Starting with models of speech
production, speech characterization, methods of analysis (transforms etc),
the authors go onto discuss pattern comparison, hidden Markov models (HMMs),
and design and implementation of speech recognition systems, right from
isolated word recognition to large vocabulary continuous speech recognition
systems. Neural networks and their use in speech recognition is also presented,
though somewhat briefly.
Rabiner was the author of the first widely-read tutorial on HMMs, so
naturally the presentation of HMMs is one of the strong points of this
textbook. The theory is developed in detail, but in an easy to follow
fashion, starting with the very basics and with plenty of helpful examples.
The implementation is discussed at great length as well, starting with
the simplest of tasks and progressing to the state-of-the-art (circa 1993).
That isn't to say that HMMs are the only good part of this book - indeed,
practically every topic, whether it be perception, transforms, vector quantization
or dynamic programming, is presented with great clarity. This book really is easy to
learn from, with numerous examples and illustrations.
The field of speech recognition is inherently multi-disciplinary in nature,
drawing upon various areas of study, including Physics, Physiology, Acoustics,
Signal Processing and Computer Science, to name but a few. The authors do a
great job of explaining all these facets, as well as the mathematics that
is an essential tool.

The only caveat is that it's now a little old (published 1993), since the
field has been growing by leaps and bounds - so while the basics remain
the same, things have changed and hence what's said here should not be
taken as the last word on the subject.
Perhaps a new edition is due, and would certainly be most welcome.
However, for an excellent, accessible introduction to this exciting field,
this is still a great choice.