Speech Recognition Over Digital Channels

Automatic speech recognition (ASR) is a very attractive means for human-machine interaction. The degree of maturity reached by speech recognition technologies during recent years allows the development of applications that use them. In particular, ASR shows an enormous potential in mobile environments, where devices such as mobile phones or PDAs are used, and for Internet Protocol (IP) applications. Speech Recognition Over Digital Channels is the first book of its kind to offer a complete system comprehension, addressing the topics of distributed and network-based speech recognition issues and standards, the concepts of speech processing and transmission, and system architectures and robustness. Describes the different client/server architectures for remote speech recognition systems, by means of which the client transmits speech parameters through a digital channel to a remote recognition server Focuses on robustness against both adverse acoustic environments (in the front-end) and bit errors/packet loss Discusses four ETSI standards for distributed speech recognition; the understanding of the standards and the technologies behind them Provides the necessary background for the comprehension of remote speech recognition technologies This book will appeal to a wide-ranging audience: engineers using speech recognition systems, researchers involved in ASR systems and those interested in processing and transmitting speech such as signal processing and communications communities. It will also be of interest to technical experts requiring an understanding of recognition over mobile and IP networks, and postgraduate students working on robust speech processing.

Antonio Peinado and Jose Segura are the authors of Speech Recognition Over Digital Channels: Robustness and Standards, published by Wiley.

Preface.

1. Introduction.

1.1 Introduction.

1.2 RSR over digital channels.

1.3 Organization of the book.

2. Speech Recognition with HMMs.

2.1 Introduction.

2.2 Some General Issues.

2.3 Analysis of Speech.

2.3.1 Preprocessing of the speech signal.

2.3.2 Linear Prediction Analysis.

2.3.3 Mel-frequency filterbanks.

2.3.4 Cepstral coeficients.

2.3.5 Other features.

2.4 Vector Quantization.

2.5 Approaches to ASR.

2.5.1 Pattern matching.

2.5.2 The statistical approach.

2.6 Hidden Markov Models.

2.6.1 Markov Processes and Hidden Markov Processes.

2.6.2 Definition of a discrete HMM.

2.6.3 The three basic problems.

2.6.4 Generalization and types of HMM modeling.

2.6.5 Simplifications to continuous HMM modeling.

2.7 Application of HMMs to speech recognition.

2.8 Model adaptation.

2.8.1 Maximum Likelihood Linear Regression (MLLR).

2.8.2 Maximum a posteriori linear regression (MAPLR).

2.9 Dealing with uncertainty.

2.9.1 Exponential weighting.

2.9.2 Bayesian Optimal classification with uncertain data.

3. Networks and Degradation.

3.1 Introduction.

3.2 Mobile and Wireless Networks.

3.2.1 Cellular structure of a mobile network.

3.2.2 Example of a mobile network architecture: GSM/GPRS.

3.2.3 Degradation in wireless networks.

3.2.4 Wireless channel models for RSR.

3.2.5 Implementation of RSR systems over mobile networks.

3.3 IP networks.

3.3.1 The TCP/IP protocol suite.

3.3.2 Degradation in IP networks.

3.3.3 Lossy packet channel models.

3.3.4 Implementation of RSR systems over packet networks.

3.4 The acoustic environment.

3.4.1 Additive noise.

3.4.2 Channel distortion.

3.4.3 A model of environment.

3.4.4 Probability distributions of noisy speech features.

4. Speech Compression and Architectures for RSR.

4.1 Introduction.

4.2 Speech Coding.

4.2.1 Waveform coders.

4.2.2 Parametric coders.

4.2.3 Pitch estimation.

4.2.4 Hybrid coders.

4.3 Recognition from Decoded Speec.

4.3.1 Effect of coding.

4.3.2 Effect of tandeming.

4.3.3 Treatment of the "coding" degradation.

4.4 Recognition from Codec Parameters.

4.4.1 Robustness of bitstream-based NSR.

4.4.2 Log-Energy computation in B-NSR systems.

4.4.3 Alternative parameter conversion.

4.5 Distributed Speech Recognition.

4.5.1 Scalar quantizers.

4.5.2 Vector quantizers and product codes.

4.5.3 Very low bit-rate PLP-based compression for DSR.

4.5.4 Transform coders.

4.5.5 Scalability in DSR systems.

4.6 Comparison between NSR and DSR.

5. Robustness Against Transmission Channel Errors.

5.1 Introduction.

5.2 Channel Coding Techniques.

5.2.1 Error detection.

5.2.2 Error correction and unequal error protection.

5.2.3 Example of channel coding for NSR: GSM-EFR.

5.2.4 Frame-level Interleaving.

5.2.5 Media-specific FEC.

5.3 Error Concealment (EC).

5.3.1 Interpolation.

5.3.2 Estimation.

5.3.3 Recognizer-based EC techniques.

5.3.4 Error concealment for NSR.

6. Front-End Processing for Robust Feature Extraction.

6.1 Introduction.

6.2 Noise reduction techniques.

6.2.1 Wiener filters.

6.2.2 Short-Time Spectral Attenuation.

6.3 Voice Activity Detection.

6.3.1 Full-band energy based VAD.

6.3.2 Statistical VAD.

6.3.3 Using long-term information.

6.4 Feature Normalization.

6.4.1 Cepstral Mean Normalization.

6.4.2 Frequency analysis of time filtered features.

6.4.3 RASTA processing.

6.4.4 Cepstral mean and variance normalization.

6.4.5 Nonlinear Feature normalization.

7. Standards for Distributed Speech Recognition.

7.1 Introduction.

7.2 Signal Preprocessing.

7.2.1 Two-stage mel-warped Wiener filtering.

7.2.2 Offset compensation and waveform processing.

7.3 Feature Extraction.

7.3.1 Computation of the basic features.

7.3.2 AFE/XAFE sampling frequency extension to 16 kHz.

7.3.3 Blind Equalization.

7.3.4 Voice Activity Detection.

7.3.5 Pitch and Class Estimation.

7.4 Feature Compression and Encoding.

7.4.1 Basic feature vector quantization.

7.4.2 Extension features quantization and encoding.

7.4.3 Bitstream/Payload format and error protection.

7.5 Feature Decoding and Postprocessing.

7.5.1 Bitstream decoding and decompression.

7.5.2 Error detection and concealment.

7.5.3 Server feature processing.

7.5.4 Pitch Tracking and Smoothing.

A. Alternative Representations of the LPC Coefficients.

B. Basic Digital Modulation Concepts.

C. Review of Channel Coding Techniques.

C.1 Media-independent FEC.

C.1.1 Linear block codes.

C.1.2 Cyclic codes.

C.1.3 Convolutional codes.

C.2 Interleaving.

Bibliography.

Index.

List of Acronyms.

What is included with this book?

The New copy of this book will include any supplemental materials advertised. Please check the title of the book to determine if it should include any access cards, study guides, lab manuals, CDs, etc.

The Used, Rental and eBook copies of this book are not guaranteed to include any supplemental materials. Typically, only the book itself is included. This is true even if the title states it includes any access cards, study guides, lab manuals, CDs, etc.

FREE SHIPPING

Rent More, Save More! Use code: BBRENTAL

5% off 1 book, 7% off 2 books, 10% off 3+ books

Speech Recognition Over Digital Channels Robustness and Standards

Summary

Author Biography

Table of Contents

Supplemental Materials