Signal Processing Examples With C64x Digital Signal Processing ...
Handbook of Signal Processing in...
Transcript of Handbook of Signal Processing in...
Handbook of SignalProcessing in Acoustics
Volume 1
Edited by
David Havelock
National Research Council, Canada
Sonoko Kuwano
Osaka University, Japan
Michael Vorlander
RWTH Aachen University, Germany
Handbook ofSignal Processingin AcousticsVolume 1
123
EditorsDavid Havelock Sonoko KuwanoNational Research Council Osaka UniversityInstitute for Microstructural Graduate School of Human
Sciences SciencesAcoustics and Signal Department of Environmental
Processing Group Psychology1200 Montreal Road 1-2 Yamadaok SuitaOttawa ON K1A 0R6 OsakaCanada Japan
Michael VorlanderRWTH Aachen UniversityInstitute of Technical AcousticsAachenGermany
ISBN: 978-0-387-77698-9 e-ISBN: 978-0-387-30441-0
Library of Congress Control Number: 2008923573
© 2008 Springer Science+Business Media, LLCAll rights reserved. This work may not be translated or copied in whole or inpart without the written permission of the publisher (Springer Science+BusinessMedia, LLC, 233 Spring Street, New York, NY 10013, USA), except for briefexcerpts in connection with reviews or scholarly analysis. Use in connection withany form of information storage and retrieval, electronic adaptation, computersoftware, or by similar or dissimilar methodology now known or hereafterdeveloped is forbidden.The use in this publication of trade names, trademarks, service marks, and similarterms, even if they are not identified as such, is not to be taken as an expressionof opinion as to whether or not they are subject to proprietary rights.Printed on acid-free paper
9 8 7 6 5 4 3 2 1
springer.com
Editor Biographies
David HavelockNational Research CouncilInstitute for Microstructural SciencesAcoustics and Signal Processing Group1200 Montreal RoadOttawa ON K1A 0R6Canada
Sonoko KuwanoOsaka UniversityGraduate School of Human SciencesDepartment of Environmental Psychology1-2 Yamadaok SuitaOsakaJapan
Michael VorlanderRWTH Aachen UniversityInstitute of Technical AcousticsAachenGermany
Editorial Board
Last Name Given Name(s) Affiliation/Address
Beauchamp James W. University of Illinois Urbana-Champaign,School of Music, Dept. of ECE,Urbana, IL, USA
Christie Douglas R. Earth Physics, Research School of EarthSciences, The Australian National University,Canberra, AUSTRALIA
Deffenbaugh Max ExxonMobil Research and EngineeringCompany, Annandale, NJ USA
Elliott Stephen J. Southampton University, Institute of Sound andVibration Research, Southampton, ENGLAND
Fastl Hugo Technische Universitaet MuenchenAG Technische Akustik, MMKMuenchen, GERMANY
Gierlich Hans Wilhelm HEAD acoustics GmbHTelecom Division, Herzogenrath, GERMANY
Guyader Jean-Louis Labratory for Vibration AcousticsVilleurbanne, FRANCE
Jacobsen Finn Technical University of Denmark,Lyngby, DENMARK
Karjalainen Matti Helsinki University of Technology, FINLAND
Kollmeier Birger Universitat Oldenburg, Oldenburg, GERMANY
O’Shaughnessy Douglas D. INRS-EMT (Telecommunications), Montreal,QC, CANADA
Riquimaroux Hiroshi Doshisha University, Department of KnowledgeEngineering & Computer Sciences, Sensory &Cognitive Neuroscience Research Laboratory,Kyotanabe, Kyoto, JAPAN
Sullivan Edmund J. EJS Consultants Portsmouth, RI, USA
viii Editorial Board
Last Name Given Name(s) Affiliation/Address
Suzuki Hideo 1-2-3-S2502, Utase, Mihama-ward, Chiba-city,JAPAN 261-0013
Taroudakis Michael University of Crete Department of Mathematics,and FORTH, Institute of Applied andComputational Mathematics, Heraklion,GREECE
Ueha Sadayuki Tokyo Institute of Technology, Director,Precision and Intelligence Lab, Yokohama,JAPAN
Verrillo Ronald T. (deceased)Institute for Sensory Research, Syracuse, NY,USA
Yamada Ichiro Aviation Environment Research Center, AirportEnvironment Improvement Foundation, Tokyo,JAPAN
Yamasaki Yoshio Waseda University, Graduate School of GlobalInformation and Telecommunication Studies,Saitama, JAPAN
Tohyama Mikio Waseda University, JAPAN
Preface
Acoustics has a special relationship with signal processing. Manyconcepts in signal processing arise naturally from our generalexperience with sound and vibration and, more than in manyother fields, acoustics is concerned with the acquisition, analysis,and synthesis of signals. Consequently, there is a rich resourceof signal processing expertise within the acoustics community.
There are many excellent reference books devoted to signalprocessing but the objective of the Handbook of Signal Processingin Acoustics is to bring together the signal processing expertisespecific to acoustics and to capture the interdisciplinary natureof signal processing within acoustics. It is also hoped that thehandbook will promote networking and the interchange of ideasbetween technical areas in acoustics.
The handbook comprises 104 Chapters organized into 17 Parts.Each Part addresses a technical area of acoustics, reflectingthe general demarcations of specialization within the acousticscommunity. An expert with broad knowledge of signal processingwithin their respective technical area was invited to act as aSection Leader for each Part of the handbook. These SectionLeaders contributed substantially to the handbook project byhelping to define the contents and scope of each chapter, findingan appropriate contributing expert author, and managing thereview and revision of material. Collectively with the Editors,they form the Editorial Board for the handbook.
Planned sections on Architectural Acoustics, NonlinearAcoustics, and Ultrasound are unfortunately omitted fromthe handbook; nevertheless, the handbook otherwise providesthorough coverage of the field of acoustics and we can hope thatpossible future editions might include these areas.
The handbook is written from the perspective of acoustics, byacousticians with signal processing expertise. Emphasis is placed
x Preface
in the description of acoustic problems and the signal processingrelated to their solutions. The reader is assumed to have basicknowledge of signal processing. Signal processing techniques aredescribed but the reader is referred elsewhere for derivations anddetails.
The authors were not required to adhere to strict standardsof style or notation, and were asked to prepare short, concise,self-sufficient chapters. This results in variations in style andnotation throughout the handbook that reflects the diversity ofperspectives within the acoustics community.
David Havelock
Acknowledgments
David Havelock gratefully acknowledges support from theInstitute for Microstructural Sciences, of the National ResearchCouncil of Canada, for making time and resources availableduring the preparation of this handbook. He also thanks John C.Burgess (University of Hawaii, retired) for the encouragement tobegin this project and expresses his gratitude to the co-Editors,Sonoko Kuwano and Michael Vorlander, who were a pleasureto work with throughout this project. The efforts of all of theSection Leaders are greatly appreciated and we thank each ofthem sincerely for their patience, perseverance, and faith thatwere required to see this project to completion. We thank eachof the authors for their contributions.
Contents
Editor Biographies v
Editorial Board vii
Preface ix
Acknowledgments xi
Contributors xvii
PART I ACOUSTIC SIGNALS AND SYSTEMS 11 Signals and Systems 3
2 Acoustic Data Acquisition 17
3 Spectral Analysis and Correlation 33
4 The FF T and Tone Identification 53
5 Measuring Transfer-Functions and Impulse Responses 65
6 Digital Sequences 87
7 Filters 107
8 Adaptive Processing 125
9 Beamforming and Wavenumber Processing 131
xiv Contents
PART II AUDITORY SYSTEM AND HEARING 14510 Anatomy, Physiology and Function of the Auditory System 147
11 Physiological Measures of Auditory Function 159
12 Auditory Processing Models 175
13 Speech Intelligibility 197
14 Signal Processing in Hearing Aids 205
PART III PSYCHOACOUSTICS 21315 Methods for Psychoacoustics in Relation
to Long-Term Sounds 215
16 Masking and Critical Bands 229
17 Aspects of Modeling Pitch Perception 241
18 Calculation of Loudness for Normaland Hearing-Impaired Listeners 251
19 Psychoacoustical Roughness 263
PART IV MUSICAL ACOUSTICS 27520 Automatic Music Transcription 277
21 Music Structure Analysis from Acoustic Signals 305
22 Computer Music Synthesis and Composition 333
23 Singing Voice Analysis, Synthesis, and Modeling 359
24 Instrument Modeling and Synthesis 375
25 Digital Waveguide Architecturesfor Virtual Musical Instruments 399
26 Modeling of Musical Instruments 419
Contents xv
PART V SPEECH 44727 Display and Analysis of Speech 449
28 Estimation of Speech Intelligibility and Quality 483
29 Gaussian Models in Automatic Speech Recognition 521
30 Speech Synthesis 557
31 Speech Coders 587
PART VI AUDIO ENGINEERING 62132 Transducer Models 623
33 Loudspeaker Design and Performance Evaluation 649
34 PA Systems for Indoor and Outdoor 669
35 Beamforming for Speech and Audio Signals 691
36 Digital Audio Recording Formats and Editing Principles 703
37 Audiovisual Interaction 731
38 Multichannel Sound Reproduction 747
39 Virtual Acoustics 761
40 Audio Restoration 773
41 Audio Effects Generation 785
42 Perceptually Based Audio Coding 797
PART VII TELECOMMUNICATIONS 81943 Speech Communication and Telephone Networks 821
44 Methods of Determining the Communicational Qualityof Speech Transmission Systems 831
xvi Contents
45 Efficient Speech Coding and TransmissionOver Noisy Channels 853
46 Echo Cancellation 883
47 Noise Reduction and Interference Cancellation 897
48 Terminals and Their Influence on Communication Quality 909
49 Networks and Their Influence on Communication Quality 915
50 Interaction of Terminals, Networksand Network Configurations 921
List of Important Abberviations 927
Contributors
Tomonari AkamatsuNational Research Institute of Fisheries Engineering, FisheriesResearch Agency of Japan, Kamisu, Ibaraki, Japan
Benoit AlcoverroCEA/DASE, BP 12, 91680 Bruyeres Le Châtel, France
Lydia AyersComputer Science Department, Hong Kong University of Scienceand Technology, Clear Water Bay, Kowloon, Hong Kong, China,e-mail: [email protected]
Juha Reinhold BackmanNokia, Espoo, Finland
Rolf BaderInstitute of Musicology, University of Hamburg, Hamburg,Germany
James W. BeauchampSchool and Music and Department of Electrical and ComputerEngineering, University of Illinois at Urbana-Champaign, 2136Music Building, 1114 West Nevada Street, Urbana, IL 61801,USA, e-mail: [email protected]
Kim BenjaminNaval Undersea Warfare Center, Newport, RI, USA
A.J. BerkhoutDelft University of Technology, Delft, The Netherlands
Jeff BilmesDepartment of Electrical Engineering, University of Washington,Seattle, WA, USA
Ole-Herman BjorNorsonic AS, Lierskogen, Norway, e-mail: [email protected]
xviii Contributors
Stanley J. Bolanowski (deceased)Institute for Sensory Research, Syracuse University, Syracuse,NY 13244, USA
Baris BozkurtIzmir Institute of Technology (IYTE), Izmir, Turkey
Nicolas BrachetProvisional Technical Secretariat, CTBTO, Vienna InternationalCentre, Vienna, Austria
Thomas BrandUniversity of Oldenburg, Oldenburg, Germany
David J. BrownGeoscience Australia, Canberra, Australia, (Provisional TechnicalSecretariat, CTBTO, Vienna International Centre, Vienna,Austria), e-mail: [email protected]
John C. BurgessDepartment of Mechanical Engineering, University of Hawaii,2540 Dole Street, Honolulu, HI 968221, USA, e-mail:[email protected]
Paola CampusProvisional Technical Secretariat, CTBTO, Vienna InternationalCentre, Vienna, Austria
Yves CansiCEA/DASE/LDG, BP12, 91680 Bruyeres-le-Châtel, France
Josef ChalupperSiemens Audiological Engineering Group, Gebbertstrasse 125,Erlangen 91058, Germany, e-mail: [email protected]
N. Ross ChapmanUniversity of Victoria, Victoria BC, Canada
Jakob Christensen-DalsgaardInstitute of Biology, University of Southern Denmark, Campusvej55, DK-5230 Odense M, Denmark, e-mail: [email protected]
Peter DanielBruel & Kjær, GmbH, Bremen, Germany
Roger B. DannenbergCarnegie Mellon University, Pittsburgh, PA, USA
Torsten DauTechnical University of Denmark, Lyngby, Denmark, e-mail:[email protected]
Contributors xix
Hans-Elias de BreeMicroflown Technologies, The Netherlands, USA
Max DeffenbaughExxonMobil Research and Engineering Company, Annandale,NJ, USA
Thierry DutoitFaculte Polytechnique de Mons, Mons, Belgium
Stephen J. ElliottInstitute of Sound and Vibration Research, Southampton, UK
Paulo A.A. EsquefNokia Institute of Technology, Rod. Torquato Tapajos, 7200,Tarumã 69048-660 Manaus-AM, Brazil, e-mail: paulo@[email protected]
Richard R. FayParmly Hearing Institute, Loyola University Chicago, 6525 N.Sheridan Rd., Chicago, IL 60626, USA, e-mail: [email protected]
Michael FehlerMassachusetts Institute of Technology, Cambridge, USA
Sandy FidellFidell Associates, Inc., Woodland Hills, CA, USA
Petr FirbasProvisional Technical Secretariat, CTBTO, Vienna InternationalCentre, Vienna, Austria (International Atomic Energy Agency,Vienna International Centre, Vienna, Austria)
Erling FrederiksenBruel & Kjær, Sound and Vibration Measurement A/S, Skods-borgvej 307, 2850 Nærum, Denmark
Milton A. GarcesInfrasound Laboratory, University of Hawaii, Manoa, HI, USA,e-mail: [email protected]
H.W. GierlichHEAD acoustics GmbH, Herzogenrath, Germany
Norbert GoertzInstitute for Digital Communications, School of Engineering andElectronics, University of Edinburgh, King’s Buildings, MayfieldRoad, Edinburgh EH9 3JL, UK
Masataka GotoNational Institute of Advanced Industrial Science andTechnology (AIST), Tokoyo, Japan
xx Contributors
Philippe GournayUniversite de Sherbrooke, Sherbrooke, QC, Canada
Jørgen HaldBruel & Kjær, Sound & Vibration Measurements A/S, Nærum,Denmark
Joe HammondUniversity of Southampton, Southampton, UK
Colin H. HansenSchool of Mechanical Engineering, University of Adelaide,Adelaide, SA 5005, Australia, e-mail: [email protected]
Uwe HansenDepartment of Physics, Indiana State University, Terre Haute,IN, USA
Sabih I. HayekDepartment of Engineering Science and Mechanics, UniversityPark, PA 16802, USA
Ulrich HeuteInstitute for Circuit and System Theory, Faculty of Engineering,Christian-Albrecht, University, Kaiserstr. 2, D-24143 Kiel,Germany
Thomas HoffmannProvisional Technical Secretariat, CTBTO, Vienna InternationalCentre, Vienna, Austria
Volker HohmannUniversity of Oldenburg, Oldenburg, Germany
Masaaki HondaWaseda University, Tokyo, Japan
Andrew B. HornerDepartment of Computer Science, Hong Kong University ofScience and Technology, Clear Water Bay, Kowloon, Hong Kong,China, e-mail: [email protected]
Adrianus J.M. HoutsmaAircrew Protection Division, U.S. Army Aeromedical ResearchLaboratory, Fort Rucker, AL 36362-0577, USA, e-mail:[email protected]
Finn JacobsenTechnical University of Denmark, Lyngby, Denmark, e-mail:[email protected]
Contributors xxi
Finn B. JensenNATO Undersea Research Centre, La Spezia, Italy, e-mail:[email protected]
Walter KellermannMultimedia Communications and Signal Processing, UniversityErlanger-Nuremberg, Erlanger, Germany, e-mail: [email protected]
Youngmoo E. KimElectrical & Computer Engineering, Drexel University, Phila-delphia, PA, USA, e-mail: [email protected]
Anssi KlapuriInstitute of Signal Processing, Tampere University of Tech-nology, Korkeakoulunkatu 1, 33720 Tampere, Finland, e-mail:[email protected]
Birger KollmeierUniversity of Oldenburg, Oldenburg, Germany
Jan Felix KrebberInstitute of Communication Acoustics, Ruhr-University Bochum,Bochum, Germany, e-mail: [email protected]
Christine E. KrohnExxonMobil Upstream Research Company, Houston, TX, USA
Sonoko KuwanoOsaka University, Osaka, Japan
Gerald C. LauchleState College, PA, USA
Walter LauriksKatholieke Universiteit Leuven, Heverlee, Belgium; Labora-torium voor Akoestick on Thermische Fysica, K.U. Leuven,Leuven, Belgium
Alexis Le PichonCEA/DASE/LDG, BP12, 91680 Bruyeres-le-Châtel, France
Philippe LeclaireUniversite de Bourgogne, Nevers, France; Laboratorium voorAkoestick on Thermische Fysica, K.U. Leuven, Leuven, Belgium
Roch LefebvreUniversite de Sherbrooke, Sherbrooke, QC, Canada
Cuiping LiDepartment of Earth and Atmospheric Science, Purdue Univer-sity, West Lafayette IN 47907, USA
xxii Contributors
Tapio LokkiDepartment of Media Technology, Helsinki University ofTechnology, Helsinki, Finland, e-mail: [email protected]
Aki Vihtori MakivirtaGenelec Oy, Iisalmi, Finland
Brian H. MarandaDefence Research and Development Canada – Atlantic, Dart-mouth, NS, Canada
James MathewsPCB Piezoelectronics Inc., San Clemente, CA, USA
Manfred MauermannUniversity of Oldenburg, Oldenburg, Germany
Walter MetznerDepartment of Physiological Science, University of California,California, USA
Yasushi MikiComputer Science, Faculty of Engineering, TakushokuUniversity, Tokyo, Japan
Ben MilnerSchool of Computing Sciences, University of East Anglia,Norwich, Norfolk, UK
Riikka MottonenDepartment of Biomedical Engineering and ComputationalScience, Helsinki University of Technology, Helsinki, Finland
Swen MullerNational Institute of Metrology, Xerem, Brazil
Seiichiro NambaOsaka University, Osaka, Japan
Ramesh NeelamaniExxonMobil Upstream Research Company, Houston, TX, USA
David E. NorrisBBN Technologies, 1300 N. 17th St., Arlington, VA 22209, USA
Robert L. NowackDepartment of Earth and Atmospheric Science, PurdueUniversity, West Lafayette, IN 47907, USA
Yasuhiro OikawaWaseda University, Tokyo, Japan
Contributors xxiii
Kazuo OkanoyaLaboratory for Biolinguistics, Brain Science Institute,Riken, 2-1 Hirosawa, Saitama 351-0198, Japan, e-mail:[email protected]
John V. OlsonGeophysical Institute, University of Alaska, Fairbanks, USA
Thorkild Find PedersenBruel & Kjær, Sound & Vibration Measurements A/S, Nærum,Denmark
Ville PulkkiDepartment of Signal Processing and Acoustics, HelsinkiUniversity of Technology, Helsinki, Finland, e-mail:[email protected]
Robert B. RandallUniversity of New South Wales, Sydney, Australia
Helmut RiedelUniversity of Oldenburg, Oldenburg, Germany
Francis RumseyInstitute of Sound Recording, University of Surrey, Guildford,UK
Masahiko SakaiOno Sokki Co., Ltd., 1-16-1 Hakusan, Midori-ku, Yokohama, 226-8507 Japan, e-mail: [email protected]
Rebecca SaltzerExxonMobil Upstream Research Company, Houston, TX, USA
Mikko SamsDepartment of Biomedical Engineering and ComputationalScience, Helsinki University of Technology, Helsinki, Finland
Lauri SaviojaDepartment of Media Technology, Helsinki University ofTechnology, Helsinki, Finland, e-mail: [email protected]
Julius O. SmithCenter for Computer Research in Music and Acoustics(CCRMA), Stanford University, Stanford, CA 94305, USA,website: http://ccrma.stanford.edu/∼jos/
Stefka StefanovaProvisional Technical Secretariat, CTBTO, Vienna InternationalCentre, Vienna, Austria
Edmund J. SullivanEJS Consultants, Portsmouth, RI, USA
xxiv Contributors
David C. SwansonThe Applied Research Laboratory, The Pennsylvania StateUniversity, Philadelphia, PA, USA
Curt A.L. SzuberlaGeophysical Institute, University of Alaska, Fairbanks, USA
Hideki TachibanaInstitute of Industrial Science, University of Tokyo, Tokyo, Japan
Yasushi TakanoDepartment of Mechanical Engineering Research Laboratory,Hitachi, Ltd., Ibaraki, Japan
Ernst TerhardtTechnical University of Munich, Munich. Germany, e-mail:[email protected]
Christine ThomasDepartment of Earth and Ocean Sciences, University ofLiverpool, Liverpool, UK
Ippei TorigoeDepartment of Mechanical Engineering and Materials, ScienceFaculty of Engineering Kumamoto University, Kumamoto, Japan
Stefan UppenkampUniversity of Oldenburg, Oldenburg, Germany
Ronald T. Verrillo (deceased)Institute for Sensory Research, Syracuse University, Syracuse,NY 13244, USA
Tuomas VirtanenInstitute of Signal Processing, Tampere University of Technology,Korkeakoulunkatu 1, 33720 Tampere, Finland, e-mail: [email protected]
Stephen VoranInstitute for Telecommunication Sciences, Boulder, CO, USA
Erhard WernerTannenweg 16, D 29693 Hademstorf, Germany
Rodney W. WhitakerLos Alamos National Laboratory, PO Box 1663, Los Alamos, NM87544, USA
Paul WhiteUniversity of Southampton, Southampton, UK
Ru-Shan WuUniversity of California, Santa Cruz, CA, USA
Contributors xxv
Xianyn WuExxonMobil Upstream Research Company, Houston, TX, USA
Ning XiangSchool of Architecture and Department of Electrical, Computer,and Systems Engineering, Rensselaer Polytechnic Institute, Troy,NY 12180, USA, e-mail: [email protected]
Ichiro YamadaAviation Environment Research Center, Airport EnvironmentImprovement Foundation, Tokyo, Japan
Kohei YamamotoKobayasi Institute of Physical Research, Tokyo, Japan
Yoshio YamasakiWaseda University, Tokyo, Japan
Nobutoshi YoshidaOno Sokki Co. Ltd., 1-16-1 Hakusan, Midori-ku, 226-8507Yokohama, Japan, e-mail: [email protected]
Udo ZolzerHelmut Schmidt University, Hamburg, Germany
PART IACOUSTIC SIGNALS
AND SYSTEMSFinn Jacobsen
Technical University of Denmark, Lyngby, Denmark
1 Signals and SystemsJoe Hammond and Paul White . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3Temporal Signal Classification • System Definition and Classification
2 Acoustic Data AcquisitionDavid C. Swanson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17Electronic Noise • Analog-to-Digital Conversion • Anti-aliasing Filters • BasicDigital Signal Calibration • Multichannel System Calibration • Data Validationand Diagnostic Techniques Summary
3 Spectral Analysis and CorrelationRobert B. Randall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33Introduction • The Fourier Transform and Variants • Practical FFT Analysis •Spectral Analysis Using Filters • Correlation Functions • Time–Frequency Analysis
4 The FFT and Tone IdentificationJohn C. Burgess . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53Introduction • The FFT (and DFT) • Leakage • Windows • Estimation ofAmplitude, Frequency, and Phase • Applications
5 Measuring Transfer-Functions and Impulse ResponsesSwen Muller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65Introduction • Measurement Methods • Excitation Signals
2 F. Jacobsen
6 Digital SequencesNing Xiang . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87Introduction • Golay Codes • Binary Maximum-Length Sequences • Fast MLSTransform (FMT) • Gold and Kasami Sequences • Legendre Sequences •Application Remarks
7 FiltersOle-Herman Bjor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107Introduction • Frequency Weightings • Octave- and Fractional-Octave-BandpassFilters • Phase Linear Filters • In-Phase/Quadrature Filters • Perceptual Masking Filters
8 Adaptive ProcessingThorkild Find Pedersen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125Introduction • The LMS Algorithm • Adaptive Applications
9 Beamforming and Wavenumber ProcessingJørgen Hald . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131Planar Nearfield Acoustical Holography • Beamforming
1Signals and SystemsJoe Hammond and Paul White
University of Southampton, Southampton, UK
Signal processing is the science of applying transformationsto measurements, to facilitate their use by an observer or acomputer, and digital signal processing (DSP) is the enablingtechnology for applications across all disciplines and sectors.
The study of sound and vibration is highly dependent on theuse of special-purpose signal analysers or software packages. Theaccessibility and convenience of DSP analysis modules and proce-dures can sometimes create a deceptive air of simplicity in oftencomplicated phenomena. It is important that practitioners, whileavailing themselves of the full range of DSP capabilities, shouldhave a clear understanding of the fundamentals of the scienceof signal processing and so be fully aware of the assumptions,implications and limitations inherent in their analysis methods.
Signal processing and analysis involves the three phases of dataacquisition, processing and interpretation (of the results of theprocessing) and, of course, all three are linked in any application.The last phase is naturally very much related to the subject underinvestigation, but the first two may be discussed independentlyof specific applications. A vast body of theory and methodologyhas been built up as a consequence of the problems raised by theneed for data analysis; this is often referred to as “signal analysis”or “time-series analysis” depending on the context [1]. The choiceof methodology is often reliant on some prior knowledge of thephenomenon being analysed. This usually relates to classifyingthe characteristics of the data and/or the way in which the datamay be modelled owing to knowledge of (or assumptions about)the way in which the data may have been generated. This section
4 J. Hammond, P. White
considers signal and system characteristics that underpin signalprocessing.
In acoustics the signal in question is generally the output of apressure transducer. Such a signal is a time history that dependson the spatial location of the transducer. We denote this by p(t,r)where t denotes the time dependence and r is the vector represen-tation for its spatial location (with Cartesian co-ordinates x, y, z).Multiple signals may be available from an array of sensors asrequired, e.g., in beamforming. For the present it is convenientto drop the spatial dependence and consider signals as evolvingwith time.
1 • Temporal Signal ClassificationThe physical phenomenon under investigation is often trans-lated by a transducer into an electrical equivalent, and a singlesignal evolving in continuous time is denoted as x�t�. In manycases, data are discrete owing to some inherent or imposedsampling procedure. In this case the data might be characterisedby a sequence of numbers. When derived from the continuoustime process x�t� we write x�n�� or xn (n = 0�1�2� � � �), wherewe have implied that the sampling interval � is constant (i.e.uniform sampling). The time histories that can occur are oftenvery complex, and it is helpful to consider signals that exhibitparticular characteristics. This allows us to relate appropriateanalysis methods to those specific types of signal.
Figure 1 illustrates a broad categorisation of signal types.A basic distinction is the designation of a signal as “random”
or “deterministic” where by “random” we mean one that is not
FIGURE 1 Classification of signals.
1 • Signals and Systems 5
exactly predictable. Very often, processes are mixed and thedemarcations shown in Figure 1 are not easily applied and conse-quently the analysis procedure to be used may not be apparent.We have included chaotic processes under both “deterministic”and “random” categories since such signals are generated froma deterministic non-linear phenomenon but nevertheless havean output with an unpredictable, random-like behaviour. Theclassification of data as being deterministic or random might bedebatable in many cases and the choice must be made on thebasis of the knowledge of the physical situation. Often signalsmay be modelled as being a mixture of both, e.g. a deterministicsignal “embedded” in unwanted random disturbances (noise).
Many of the classes defined above, such as the stationary andperiodic signals, are mathematical constructs to which no real-world signal can belong. However, these classes do provide onewith a set of models which, in many cases, provide good approx-imations to measured processes and they suggest a suitableanalysis framework.
A brief note for each of the categories is given below –continuous time is used throughout. An analysis method appro-priate for each is also noted.
1.1 Periodic A signal is periodic with period Tp if x�t� = x�t+Tp�.Such a signal can be represented in the frequency domain as a
Fourier series:
x �t� = a0 +�∑
k=1
ak cos �2�kf0t�+bk sin �2�kf0t� =�∑
k=�cke2�ikf0t� (1)
The feature of this is that the frequencies in representation arethe fundamental, f0 = 1/Tp, and multiples (harmonics) (plus a d.c.term), and the coefficients, ak and bk (or ck), give the amplitudeand phase of the components.
Figure 2a shows the time-series of a small section of voicedspeech, the vowel /e/. The approximately periodic character ofthis signal is evident and suggests that it could be gainfullyanalysed by computing its Fourier series representation. ThisFourier series is shown in Figure 2b, which has been computedbased on the first period of the signal shown in Figure 2a. Oneshould appreciate that the assumption of periodicity does nothold exactly for this example, as is commonly the case.
6 J. Hammond, P. White
FIGURE 2 Example of voiced speech: (a) time-series, (b) Fouriercoefficients.
1 • Signals and Systems 7
1.2 Transient A transient signal is one which is essentially localised, i.e. a signalthat has a finite duration. Such a signal has a Fourier integralrepresentation:
x �t� =�∫
−�X �f �e2�iftdf� (2)
This differs from the periodic case in that the frequency rangebecomes a continuum and amplitudes “in a band” are X�f�df soX�f� is now an amplitude density.
Figure 3a shows an example of the time-series of an unvoicedspeech segment, in this case the sound /th/. The lack of a periodicstructure in this signal is apparent. It can be considered as atransient signal and analysed by employing the Fourier transformdefined in (2), which is shown in Figure 3b.
1.3 AlmostPeriodic
This could be regarded as a process where Tp in (1) varies withtime. There is no natural generalisation of (1) for this.
An alternative way of constructing a simple almost periodicprocess is, e.g.,
x �t� = sin �2�t�+ sin(2�
√2t)
� (3)
This process whilst being the sum of two periodic signals is itselfnot periodic because the ratio of the two frequency componentsis not a rational number. In this case using the Fourier integral(2) is appropriate.
Figure 4a shows the time-series of a segment of a sustainedpiano note. These notes contain discrete frequencies that areapproximately harmonically related. The inharmonicities presentin piano string vibrations mean that such signals should beregarded as almost periodic and so the use of the Fouriertransform (2) represents a suitable tool for analysis; the resultsof such an analysis are shown in Figure 4b.
1.4 ChaoticProcess
The range of phenomena that lead to chaotic dynamics iswide, but a feature is that the describing equation is decep-tively simple in its deterministic form, but generates intricatesignals. An example is a second-order non-linear system drivenby a sinusoidal excitation. The methodology of analysis usestopological feature analysis, by which one attempts to reveal theunderlying simple generation mechanism.
8 J. Hammond, P. White
FIGURE 3 Example of unvoiced speech: (a) time-series, (b) magnitude ofthe Fourier transform.