Speech Recognition: Theory and C++ Implementation

It's very good.

This book is composed of two parts, theory and implementation. if you only read its theoretical part, it is ok with many details missing. it is not clearly written. however, if you study its C++ code, you would get all you want on recognition system. I spent 3-8 hours everyday for 4 months going through its code line by line. The C++ code (30,000 lines in total) is very well written but without comments. Many times, I need to figure out things not written in the book. I once spent 1 week on 200 lines of code. However, After 4 months, I truely understand the system. You will find this book useful only if you really spend time covering its C++ code line by line. If you want theory only, goto read other books. I rate 2 star for its theory and 5 star for its implementation. [website]

Speech Recognition: Theory and C++ Implementation This book is empty and trivial, and there is no companion CD_ROM as mention inside the book, the C++ code in the internet is poorly documented and not working,finaly don't puchase this book.

I appreciated the balance between theory and implementation in the book. Also the content covers the most important topics. It is unfortunate that the book contains numerous typo's and confusing choices of symbols. The errors are often right in the most critical places too. The explanation of the theory of HMM's for example. 0 and 1 are chosen to represent both white and black balls and 2 different urns all in the same diagram. When trying to sort out which is which the reader will be further confused by blatant errors where a 0 should be a 1. I am afraid many new readers will find frustration on the theory sections. The choice of C++ and inclusion of a CD-ROM with full source is a nice touch however. Just be aware that the code is not geared for real-time recognition.

For studying the speech recognition subject this is not the right book to buy, It is hard to understand the theory using this book. The c++ code works but there should be more remarks to make it easier to the readers to get along. There is a free toolkit to download from the Internet named HTK that contains full C code and a free book to the same theory so actually it is a waste of money to buy this book !!!

I did my PhD in signal processing but have very few background on automatic speech recognition (ASR). I was searching for a book that can guide me quickly to at least important methods in ASR, but at the same time have a clear instruction on implementing an ASR system. So in short, I want to code (in C) an ASR but with understanding hor it works. This book does a perfect job on this. However, the writing is often confusing. My friend (a researcher, native English speaker) found it hard to understand what they write. For me, I often apply the strategy: try to understand the general concept in the text. Equations sometimes have typo, otherwise, they are in simplified form and easy to follow.