Fills a market gap for an up-to-date text. Ex.___
Demonstrates the latest developments in the application of speech modification, speech enhancement, speech coding, and speaker recognition. Ex.___
Illustrates for the student that some techniques have different solution paths depending on the problem. Ex.___
Presents the most intensive set of examples and exercises available in a speech processing text. Ex.___
Provides important hands-on experience with speech signals and speech signal processing, essential as a learning tool. Ex.___
Illustrate the speech processing concepts. Ex.___
Here you can download a set of audio demonstrations collected originally for Thomas Quatieri's MIT course Digital Speech Processing, which he taught six times over the past decade. The audio demonstrations serve to illustrate principles, examples, and exercises that are introduced throughout the text and do not necessarily represent the state-of-the-art in a particular area or of a particular technique given the continued progress being made in this field. The demonstrations are by no means complete, representing a (non-uniform) sparse sampling of the material.
Each chapter has an associated folder of audio demonstrations. Each chapter folder contains a readme file that describes its contents and audio files in both pcm format (16-bit integer) and wav format. In addition, there occassionally appear figures (e.g., comparative spectrograms of speech file segments) to illustrate a concept and/or supplement the text.
To illustrate, the directory structure for Chapter 9 is given by:
In addition to the chapter folders, there is a folder called "scripts" that contains in-house MATLAB and Unix routines that were used in creating the pcm, wav, and figure files.
Essential principles, practical examples, current applications, and leading-edge research.
In this book, Thomas F. Quatieri presents the field's most intensive, up-to-date tutorial and reference on discrete-time speech signal processing. Building on his MIT graduate course, he introduces key principles, essential applications, and state-of-the-art research, and he identifies limitations that point the way to new research opportunities.
Quatieri provides an excellent balance of theory and application, beginning with a complete framework for understanding discrete-time speech signal processing. Along the way, he presents important advances never before covered in a speech signal processing text book, including sinusoidal speech processing, advanced time-frequency analysis, and nonlinear aeroacoustic speech production modeling. Coverage includes:
The book's in-depth applications coverage includes speech coding, enhancement, and modification; speaker recognition; noise reduction; signal restoration; dynamic range compression, and more. Principles of Discrete-Time Speech Processing also contains an exceptionally complete series of examples and Matlab exercises, all carefully integrated into the book's coverage of theory and applications.
Here you can download functions, scripts, workspaces, and data required by the MATLAB exercises in the book. Note that the speech files given are often specific cases of a certain class of speech signal, i.e., the waveforms often can be replaced by other examples with similar characteristics that serve the instructive purpose of the exercise.
Each chapter has its own directory structure. As an example, the directory for Chapter 11 consists of:
In addition, readme files occasionally appear within the subdirectories. These readme files are in text format and describe or clarify contents of the subdirectory or a particular MATLAB exercise.
Speech files in the Data directory are in both pcm and wav format (but are also included in the given workspaces). The sampling rate of each speech file is indicated in the file name. The Data directory for Chapter 11, for example, contains the required speech files:
If the designated sampling is not supported by a desired platform, then listening to the files entails a change in sampling rate (e.g., using MATLAB decimation and interpolation functions) for the particular platform.
To obtain the MATLAB excercises, you can download the files for all chapters in a single archive (4.1 MB, zip).
See the 'Audio & Video" tab for downloadable audio demonstration files.
(NOTE: Each chapter begins with an introduction and concludes with a Summary, Exercises and Bibliography.)
Discrete-Time Speech Signal Processing. The Speech Communication Pathway. Analysis/Synthesis Based on Speech Production and Perception. Applications. Outline of Book.
2. A Discrete-Time Signal Processing Framework.
Discrete-Time Signals. Discrete-Time Systems. Discrete-Time Fourier Transform. Uncertainty Principle. z-Transform. LTI Systems in the Frequency Domain. Properties of LTI Systems. Time-Varying Systems. Discrete-Fourier Transform. Conversion of Continuous Signals and Systems to Discrete Time.
3. Production and Classification of Speech Sounds.
Anatomy and Physiology of Speech Production. Spectrographic Analysis of Speech. Categorization of Speech Sounds. Prosody: The Melody of Speech. Speech Perception.
4. Acoustics of Speech Production.
Physics of Sound. Uniform Tube Model. A Discrete-Time Model Based on Tube Concatenation. Vocal Fold/Vocal Tract Interaction.
5. Analysis and Synthesis of Pole-Zero Speech Models.
Time-Dependent Processing. All-Pole Modeling of Deterministic Signals. Linear Prediction Analysis of Stochastic Speech Sounds. Criterion of “Goodness”. Synthesis Based on All-Pole Modeling. Pole-Zero Estimation. Decomposition of the Glottal Flow Derivative. Appendix 5.A: Properties of Stochastic Processes.
Random Processes. Ensemble Averages. Stationary Random Process. Time Averages. Power Density Spectrum. Appendix 5.B: Derivation of the Lattice Filter in Linear Prediction Analysis.
6. Homomorphic Signal Processing.
Concept. Homomorphic Systems for Convolution. Complex Cepstrum of Speech-Like Sequences. Spectral Root Homomorphic Filtering. Short-Time Homomorphic Analysis of Periodic Sequences. Short-Time Speech Analysis. Analysis/Synthesis Structures. Contrasting Linear Prediction and Homomorphic Filtering. 7. Short-Time Fourier Transform Analysis and Synthesis.
Short-Time Analysis. Short-Time Synthesis. Short-Time Fourier Transform Magnitude. Signal Estimation from the Modified STFT or STFTM. Time-Scale Modification and Enhancement of Speech. Appendix 7.A: FBS Method with Multiplicative Modification.
8. Filter-Bank Analysis/Synthesis.
Revisiting the FBS Method. Phase Vocoder. Phase Coherence in the Phase Vocoder. Constant-Q Analysis/Synthesis. Auditory Modeling. 9. Sinusoidal Analysis/Synthesis.
Sinusoidal Speech Model. Estimation of Sinewave Parameters. Synthesis. Source/Filter Phase Model. Additive Deterministic-Stochastic Model. Appendix 9.A: Derivation of the Sinewave Model.
Appendix 9.B: Derivation of Optimal Cubic Phase Parameters.
10. Frequency-Domain Pitch Estimation.
A Correlation-Based Pitch Estimator. Pitch Estimation Based on a “Comb Filter<170. Pitch Estimation Based on a Harmonic Sinewave Model. Glottal Pulse Onset Estimation. Multi-Band Pitch and Voicing Estimation. 11. Nonlinear Measurement and Modeling Techniques.
The STFT and Wavelet Transform Revisited. Bilinear Time-Frequency Distributions. Aeroacoustic Flow in the Vocal Tract. Instantaneous Teager Energy Operator. 12. Speech Coding.
Statistical Models of Speech. Scaler Quantization. Vector Quantization (VQ). Frequency-Domain Coding. Model-Based Coding. LPC Residual Coding. 13. Speech Enhancement.
Introduction. Preliminaries. Wiener Filtering. Model-Based Processing. Enhancement Based on Auditory Masking. Appendix 13.A: Stochastic-Theoretic parameter Estimation.
14. Speaker Recognition.
Introduction. Spectral Features for Speaker Recognition. Speaker Recognition Algorithms. Non-Spectral Features in Speaker Recognition. Signal Enhancement for the Mismatched Condition. Speaker Recognition from Coded Speech. Appendix 14.A: Expectation-Maximization (EM) Estimation.
Glossary.Speech Signal Processing.Units.Databases.Index.About the Author.
This text is in part an outgrowth of my MIT graduate course Digital Speech Signal Processing, which I have taught since the Fall of 1990, and in part a result of my research at MIT Lincoln Laboratory. As such, principles are never too distant from practice; theory is often followed by applications, both past and present. This text is also an outgrowth of my childhood wonder in the blending of signal and symbol processing, sound, and technology. I first felt this fascination in communicating with two cans coupled by twine, in playing with a toy Morse code, and in adventuring through old ham radio equipment in my family's basement. My goals in this book are to provide an intensive tutorial on the principles of discrete-time speech signal processing, to describe the state-of-the-art in speech signal processing research and its applications, and to pass on to the reader my continued wonder for this rapidly evolving field.
The text consists of fourteen chapters that are outlined in detail in Chapter 1. The "theory" component of the book falls within Chapters 2-11, while Chapters 12-14 consist primarily of the application areas of speech coding and enhancement, and speaker recognition. Other applications are introduced throughout Chapters 2-11, such as speech modification, noise reduction, signal restoration, and dynamic range compression. A broader range of topics that include speech and language recognition is not covered; to do so would result in a survey book that does not fill the current need in this field. The style of the text is to show not only when speech modeling and processing methods succeed, but also to describe limitations of the methods. This style makes the reader question established ideas and reveals where advancement is needed. An important tenet in this book is that anomaly in observation is crucial for advancement; as reflected by the late philosopher Thomas Kuhn: "Discovery commences with the awareness of anomaly, i.e., with the recognition that nature has somehow violated the paradigm-induced expectations that govern normal science."1
The text body is strongly supplemented with examples and exercises. Each exercise set contains a number of MATLAB problems that provide hands-on experience with speech signals and processing methods. Scripts, workspaces, and signals, required for the MATLAB exercises, are located on the Prentice Hall companion website (
http://www.phptr.com/quatieri/). Also on this website are audio demonstrations that illustrate a variety of principles and applications from each chapter, including time-scale modification of the phrase "as time goes by" shown on the front cover of this book. The book is structured so that application areas that are not covered as separate topics are either presented as examples or exercises, e.g., speaker separation by sinusoidal modeling and restoration of old acoustic recordings by homomorphic processing. In my MIT speech processing course, I found this approach to be very effective, especially since such examples and exercises are fascinating demonstrations of the theory and can provide a glimpse of state-of-the-art applications.
The book is also structured so that topics can be covered on different levels of depth and breadth. For example, a one-semester course on discrete-time speech signal processing could be taught with an emphasis on fundamentals using Chapters 2-9. To focus on the speech coding application, one can include Chapter 12, but also other applications as examples and exercises. In a two-semester course, greater depth could be given to fundamentals in the first semester, using Chapters 2-9. In the second semester, a focus could then be given to advanced theories and applications of Chapters 10-14, with supplementary material on speech recognition.
Select from the pages/chapters below to download a PDF(s) of replacement pages: