This book reviews past and present work on discriminative and hierarchical models for both acoustic and language modeling. It also analyzes the research direction and trends towards establishing future-generation speech recognition.

This book provides a comprehensive overview of the recent advancement in the field of automatic speech recognition with a focus on deep learning models including deep neural networks and many of their variants. This is the first automatic speech recognition book dedicated to the deep learning approach. In addition to the rigorous mathematical treatment of the subject, the book also presents insights and theoretical foundation of a series of highly successful deep learning models.

Presents important theoretical foundation and practical considerations of using a wide range of deep learning models and methods for automatic speech recognition Reviews past and present work (up to the fall of year 2014) on most impactful work based on deep learning for acoustic modeling in speech recognition Goes deeply into rigorous mathematical and technical descriptions of deep learning methods successful for speech recognition and related areas of applications Analyzes research directions and trends towards establishing future-generation speech recognition based on extending the current deep learning models Includes supplementary material: sn.pub/extras

Inhalt
Section 1: Automatic speech recognition: Background.- Feature extraction: basic frontend.- Acoustic model: Gaussian mixture hidden Markov model.- Language model: stochastic N-gram.- Historical reviews of speech recognition research: 1st, 2nd, 3rd, 3.5th, and 4th generations.- Section 2: Advanced feature extraction and transformation.- Unsupervised feature extraction.- Discriminative feature transformation.- Section 3: Advanced acoustic modeling.- Conditional random field (CRF) and hidden conditional random field (HCRF).- Deep-Structured CRF.- Semi-Markov conditional random field.- Deep stacking models.- Deep neural network hidden Markov hybrid model.- Section 4: Advanced language modeling.- Discriminative Language model.- Log-linear language model.- Neural network language model.