The International Series in Engineering and Computer Science - 355: Automatic Speech and Speaker Recognition

More about the book

Research in automatic speech and speaker recognition has significantly advanced over the past two decades, driven by improvements in signal processing, algorithms, architectures, and hardware. Key developments include the adoption of a statistical pattern recognition approach, the use of hidden Markov modeling to capture both spectral and temporal variations in speech, and training these models with extensive speech data from diverse speakers. Additionally, speech and language knowledge sources are organized into structural finite state networks, and dynamic programming-based heuristic search methods are employed to identify the best word sequences corresponding to spoken utterances. This volume consolidates crucial topics in speech and speaker recognition that remain underrepresented in existing literature. Although not explicitly divided, the content can be categorized into five parts: Chapters 1-2 provide technology overviews; Chapters 3-12 focus on acoustic modeling of speech units and lexical modeling of words; Chapters 13-15 explore flexibility and robustness; Chapters 16-18 address theoretical and practical search issues; and Chapters 19-20 present examples of algorithms and implementation for recognition systems. This book serves as a reference for speech researchers and graduate students interested in pursuing research in this field and may also be utilized as a textbook for advanced courses on the subject.