TY  - BOOK
AU  - Morgan,David P.
AU  - Scofield,Christopher L.
TI  - Neural networks and speech processing
SN  - 0792391446 (alk. paper)
AV  - QA76.87 .M67 1991
U1  - 006.3 MOR S 23
PY  - 1991///
CY  - Boston
PB  - Kluwer Academic Publishers
KW  - Neural networks (Computer science)
KW  - Speech processing systems
N1  - Includes bibliographical references (p. 359-382) and index; 1 Introduction.- 1.1 Motivation.- 1.2 A Few Words on Speech Recognition.- 1.3 A Few Words on Neural Networks.- 1.4 Contents.- 2 The Mammalian Auditory System.- 2.1 Introduction to Auditory Processing.- 2.2 The Anatomy and Physiology of Neurons.- 2.3 Neuroanatomy of the Auditory System.- 2.3.1 The Ear.- 2.3.2 The Cochlea.- 2.3.3 The Eighth Nerve.- 2.3.4 The Cochlear Nucleus.- 2.3.5 The Superior Olivary Complex.- 2.3.6 The Inferior Colliculus.- 2.3.7 The Medial Geniculate Nucleus.- 2.3.8 The Auditory Cortex.- 2.4 Recurrent Connectivity in the Auditory Pathway.- 2.5 Summary.- 3 An Artificial Neural Network Primer.- 3.1 A Neural Network Primer for Speech Scientists.- 3.2 Elements of Artificial Neural Networks.- 3.2.1 Similarity Measures and Activation Functions.- 3.2.2 Networks and Mappings.- 3.3 Learning in Neural Networks.- 3.4 Supervised Learning.- 3.4.1 The Perceptron and Gradient-Descent Learning.- 3.4.2 Associative Memories.- 3.4.3 The Hopfield Network.- 3.5 Multi-Layer Networks.- 3.5.1 The Restricted Coulomb Energy Network.- 3.5.2 The Backward Error Propagation Network.- 3.5.3 The Charge Clustering Network.- 3.5.4 Recurrent Back Propagation.- 3.6 Unsupervised Learning.- 3.6.1 The BCM Network.- 3.6.2 The Kohonen Feature Map.- 3.7 Summary.- 4 A Speech Technology Primer.- 4.1 A Speech Primer for Neural Scientists.- 4.2 Human Speech Production/Perception.- 4.2.1 Information in the Speech Signal.- 4.3 ASR Technology.- 4.3.1 A General Speech Recognition Model.- 4.4 Signal Processing and Feature Extraction.- 4.4.1 Linear Predictive Coding.- 4.4.2 Feature Extraction and Modeling.- 4.4.3 Vector Quantization.- 4.5 Time Alignment and Pattern Matching.- 4.5.1 Dynamic Time Warping.- 4.5.2 Hidden Markov Models.- 4.5.3 Pronunciation Network Word Models.- 4.6 Language Models.- 4.6.1 Parsers.- 4.6.2 Statistical Models.- 4.7 Summary.- 5 Methods in Neural Network Applications.- 5.1 The Allure of Neural Networks for Speech Processing.- 5.2 The Computational Properties of ANNs.- 5.2.1 Computability and Network Size.- 5.3 ANN Limitations: The Scaling Problem.- 5.3.1 The Scaling of Learning.- 5.3.2 The Scaling of Generalization.- 5.4 Structured ANN Solutions.- 5.4.1 Hierarchical Modules.- 5.4.2 Hybrid Systems.- 5.4.3 Multiple Neural Network Systems.- 5.4.4 Integrating Neural Speech Modules.- 5.5 Summary.- 6 Signal Processing and Feature Extraction.- 6.1 The Importance of Signal Representations.- 6.2 The Signal Processing Problem Domain.- 6.3 Biologically Motivated Signal Processing.- 6.3.1 Review of Speech Representation in the Auditory Nerve.- 6.3.2 The Silicon Cochlea and Temporal-Place Representations for ASR.- 6.3.3 The Role of Automatic Gain Control in Noisy Environments.- 6.4 ANNs for Conventional Signal Processing.- 6.4.1 Adaptive Filtering.- 6.4.2 A Noise Reduction Network.- 6.5 Feature Representations.- 6.5.1 Unsupervised Feature Extraction for Phoneme Classification.- 6.5.2 Feature Maps.- 6.6 Summary.- 7 Time Alignment and Pattern Matching.- 7.1 Modeling Spectro-Temporal Structure.- 7.2 Time Normalization Via Pre-Processing.- 7.2.1 Interpolation and Decimation Techniques.- 7.2.2 Feature-Set Transformations.- 7.3 The Dynamic Programming Neural Network.- 7.3.1 The DPNN Architecture.- 7.3.2 The Time Warping Structure.- 7.3.3 The DPNN Training Procedure.- 7.3.4 Application to Speaker-Independent Digit Recognition.- 7.4 HMM Motivated Networks.- 7.4.1 The Viterbi Network.- 7.4.2 The HMM Network.- 7.5 Recurrent Networks for Temporal Modeling.- 7.5.1 The Temporal Flow Model.- 7.5.2 Temporal Flow Experiments.- 7.6 The Time Delay Neural Network.- 7.6.1 The TDNN Temporal Architecture.- 7.6.2 TDNN Training.- 7.6.3 Application to Phoneme Classification.- 7.6.4 Interpreting the TDNN Spectro-Temporal Representation.- 7.6.5 Phoneme Classification Summary.- 7.6.6 TDNNs for Word Discrimination.- 7.7 Summary.- 8 Natural Language Processing.- 8.1 The Importance of Language Processing.- 8.2 Syntactic Models.- 8.2.1 NETgrams: An ANN Word Category Predictor.- 8.2.2 An ANN for Word Category Disambiguation.- 8.2.3 Recurrent Networks and Formal Languages.- 8.3 Semantic Models.- 8.3.1 Pronoun Reference ANNs.- 8.4 Knowledge Representation.- 8.4.1 Knowledge Representation in a Hopfield Network.- 8.5 Summary.- 9 ANN Keyword Recognition.- 9.1 Keyword Spotting.- 9.2 The Primary KWS System.- 9.2.1 Experimental Data.- 9.3 DUR Experiments.- 9.3.1 Selecting a Fixed-Length Feature Representation.- 9.3.2 Single and Multiple Networks.- 9.3.3 Experiments with Hybrid Systems.- 9.4 Secondary Processing Experiments.- 9.4.1 The Pattern Matching Approach.- 9.4.2 An Investigation of Temporal Models.- 9.5 Summary.- 10 Neural Networks and Speech Processing.- 10.1 Speech Processing Applications.- 10.1.1 Speech Synthesis.- 10.1.2 Speech Coding.- 10.1.3 Speaker Separation.- 10.1.4 Speech Enhancement.- 10.1.5 Speaker Verification/Identification.- 10.1.6 Language Identification.- 10.1.7 Keyword/Keyphrase Spotting.- 10.2 Summary of Efforts in ASR.- 10.2.1 The Past: Institutions Involved in ASR.- 10.2.2 The Current Status of ANNs in ASR.- 10.2.3 The Future: Challenges and Goals.- 10.3 Concluding Remarks
ER  -