【图书】AdvancedDigital SignalProcessing，pdf有目录，绝对超值！！！

forestmeng

1 Introduction 1
1.1Signals,NoiseandInformation 1
1.2SignalProcessingMethods 3
1.2.1Transform-BasedSignalProcessing 3
1.2.2Source-FilterModel-BasedSignalProcessing 5
1.2.3BayesianStatisticalModel-BasedSignalProcessing 5
1.2.4NeuralNetworks 6
1.3ApplicationsofDigitalSignalProcessing 6
1.3.1DigitalWatermarking 6
1.3.2Bio-medical,MIMO,SignalProcessing 8
1.3.3EchoCancellation 10
1.3.4AdaptiveNoiseCancellation 12
1.3.5AdaptiveNoiseReduction 12
1.3.6BlindChannelEqualisation 13
1.3.7SignalClassificationandPatternRecognition 13
1.3.8LinearPredictionModellingofSpeech 15
1.3.9DigitalCodingofAudioSignals 16
1.3.10DetectionofSignalsinNoise 17
1.3.11DirectionalReceptionofWaves:Beam-forming 18
1.3.12Space-TimeSignalProcessing 20
1.3.13DolbyNoiseReduction 20
1.3.14RadarSignalProcessingopplerFrequencyShift 21
1.4AReviewofSamplingandQuantisation 22
1.4.1AdvantagesofDigitalFormat 24
1.4.2DigitalSignalsStoredandTransmittedinAnalogueFormat 25
1.4.3TheEffectofDigitisationonSignalBandwidth 25
1.4.4SamplingaContinuous-TimeSignal 25
1.4.5AliasingDistortion 27
1.4.6NyquistSamplingTheorem 27
viii Contents
1.4.7Quantisation 28
1.4.8Non-LinearQuantisation,Companding 30
1.5Summary 32
Bibliography 32
2 NoiseandDistortion 35
2.1Introduction 35
2.1.1DifferentClassesofNoiseSourcesandDistortions 36
2.1.2DifferentClassesandspectral/TemporalShapesofNoise 37
2.2WhiteNoise 37
2.2.1Band-LimitedWhiteNoise 38
2.3ColouredNoiseinkNoiseandBrownNoise 39
2.4ImpulsiveandClickNoise 39
2.5TransientNoisePulses 41
2.6ThermalNoise 41
2.7ShotNoise 42
2.8Flicker(I/f)Noise 43
2.9BurstNoise 44
2.10Electromagnetic(Radio)Noise 45
2.10.1NaturalSourcesofRadiationofElectromagneticNoise 45
2.10.2Man-madeSourcesofRadiationofElectromagneticNoise 45
2.11ChannelDistortions 46
2.12EchoandMulti-pathReflections 47
2.13ModellingNoise 47
2.13.1FrequencyAnalysisandCharacterisationofNoise 47
2.13.2AdditiveWhiteGaussianNoiseModel(AWGN) 48
2.13.3HiddenMarkovModelandGaussianMixtureModelsforNoise 49
Bibliography 50
3 InformationTheoryandProbabilityModels 51
3.1IntroductionrobabilityandInformationModels 52
3.2RandomProcesses 53
3.2.1Information-bearingRandomSignalsvsDeterministicSignals 53
3.2.2Pseudo-RandomNumberGenerators(PRNG) 55
3.2.3StochasticandRandomProcesses 56
3.2.4TheSpaceofVariationsofaRandomProcess 56
3.3ProbabilityModelsofRandomSignals 57
3.3.1ProbabilityasaNumericalMappingofBelief 57
3.3.2TheChoiceofOneandZeroastheLimitsofProbability 57
3.3.3Discrete,ContinuousandFinite-StateProbabilityModels 58
3.3.4RandomVariablesandRandomProcesses 58
3.3.5ProbabilityandRandomVariables–TheSpaceandSubspacesofa
Variable 58
3.3.6ProbabilityMassFunction–DiscreteRandomVariables 60
3.3.7Bayes’Rule 60
3.3.8ProbabilityDensityFunction–ContinuousRandomVariables 61
3.3.9ProbabilityDensityFunctionsofContinuousRandomProcesses 62
3.3.10Histograms–ModelsofProbability 63
3.4InformationModels 64
3.4.1Entropy:AMeasureofInformationandUncertainty 65
3.4.2MutualInformation 68
Contents ix
3.4.3EntropyCoding–VariableLengthCodes 69
3.4.4HuffmanCoding 70
3.5StationaryandNon-StationaryRandomProcesses 73
3.5.1Strict-SenseStationaryProcesses 75
3.5.2Wide-SenseStationaryProcesses 75
3.5.3Non-StationaryProcesses 76
3.6Statistics(ExpectedValues)ofaRandomProcess 76
3.6.1CentralMoments 77
3.6.1.1Cumulants 77
3.6.2TheMean(orAverage)Value 77
3.6.3Correlation,SimilarityandDependency 78
3.6.4Autocovariance 81
3.6.5PowerSpectralDensity 81
3.6.6JointStatisticalAveragesofTwoRandomProcesses 83
3.6.7Cross-CorrelationandCross-Covariance 83
3.6.8Cross-PowerSpectralDensityandCoherence 84
3.6.9ErgodicProcessesandTime-AveragedStatistics 85
3.6.10Mean-ErgodicProcesses 85
3.6.11Correlation-ErgodicProcesses 86
3.7SomeUsefulPracticalClassesofRandomProcesses 87
3.7.1Gaussian(Normal)Process 87
3.7.2MultivariateGaussianProcess 88
3.7.3GaussianMixtureProcess 89
3.7.4Binary-StateGaussianProcess 90
3.7.5PoissonProcess–CountingProcess 91
3.7.6ShotNoise 92
3.7.7Poisson–GaussianModelforCluttersandImpulsiveNoise 93
3.7.8MarkovProcesses 94
3.7.9MarkovChainProcesses 95
3.7.10HomogeneousandInhomogeneousMarkovChains 96
3.7.11GammaProbabilityDistribution 96
3.7.12RayleighProbabilityDistribution 97
3.7.13ChiDistribution 97
3.7.14LaplacianProbabilityDistribution 98
3.8TransformationofaRandomProcess 98
3.8.1MonotonicTransformationofRandomProcesses 99
3.8.2Many-to-OneMappingofRandomSignals 101
3.9SearchEngines:CitationRanking 103
3.9.1CitationRankinginWebPageRankCalculation 104
3.10Summary 104
Bibliography 105
4 BayesianInference 107
4.1BayesianEstimationTheory:BasicDefinitions 108
4.1.1Bayes’Theorem 109
4.1.2ElementsofBayesianInference 109
4.1.3DynamicandProbabilityModelsinEstimation 110
4.1.4ParameterSpaceandSignalSpace 111
4.1.5ParameterEstimationandSignalRestoration 111
4.1.6PerformanceMeasuresandDesirablePropertiesofEstimators 112
4.1.7PriorandPosteriorSpacesandDistributions 114
x Contents
4.2BayesianEstimation 117
4.2.1MaximumAPosterioriEstimation 117
4.2.2Maximum-Likelihood(ML)Estimation 118
4.2.3MinimumMeanSquareErrorEstimation 121
4.2.4MinimumMeanAbsoluteValueofErrorEstimation 122
4.2.5EquivalenceoftheMAP,ML,MMSEandMAVEEstimatesforGaussian
ProcesseswithUniformDistributedParameters 123
4.2.6InfluenceofthePrioronEstimationBiasandVariance 123
4.2.7RelativeImportanceofthePriorandtheObservation 126
4.3Expectation-Maximisation(EM)Method 128
4.3.1CompleteandIncompleteData 128
4.3.2MaximisationofExpectationoftheLikelihoodFunction 129
4.3.3DerivationandConvergenceoftheEMAlgorithm 130
4.4Cramer–RaoBoundontheMinimumEstimatorVariance 131
4.4.1Cramer–RaoBoundforRandomParameters 133
4.4.2Cramer–RaoBoundforaVectorParameter 133
4.5DesignofGaussianMixtureModels(GMMs) 134
4.5.1EMEstimationofGaussianMixtureModel 134
4.6BayesianClassification 136
4.6.1BinaryClassification 137
4.6.2ClassificationError 139
4.6.3BayesianClassificationofDiscrete-ValuedParameters 139
4.6.4MaximumAPosterioriClassification 140
4.6.5Maximum-LikelihoodClassification 140
4.6.6MinimumMeanSquareErrorClassification 140
4.6.7BayesianClassificationofFiniteStateProcesses 141
4.6.8BayesianEstimationoftheMostLikelyStateSequence 142
4.7ModellingtheSpaceofaRandomProcess 143
4.7.1VectorQuantisationofaRandomProcess 143
4.7.2VectorQuantisationusingGaussianModelsofClusters 143
4.7.3DesignofaVectorQuantiser:K-MeansClustering 144
4.8Summary 145
Bibliography 146
5 HiddenMarkovModels 147
5.1StatisticalModelsforNon-StationaryProcesses 147
5.2HiddenMarkovModels 149
5.2.1ComparisonofMarkovandHiddenMarkovModels 149
5.2.1.1Observable-StateMarkovProcess 149
5.2.1.2Hidden-StateMarkovProcess 149
5.2.2APhysicalInterpretation:HMMsofSpeech 151
5.2.3HiddenMarkovModelasaBayesianModel 152
5.2.4ParametersofaHiddenMarkovModel 152
5.2.5StateObservationProbabilityModels 153
5.2.6StateTransitionProbabilities 154
5.2.7State–TimeTrellisDiagram 154
5.3TrainingHiddenMarkovModels 155
5.3.1Forward–BackwardProbabilityComputation 156
5.3.2Baum–WelchModelRe-estimation 157
5.3.3TrainingHMMswithDiscreteDensityObservationModels 158
Contents xi
5.3.4HMMswithContinuousDensityObservationModels 159
5.3.5HMMswithGaussianMixturepdfs 160
5.4DecodingSignalsUsingHiddenMarkovModels 161
5.4.1ViterbiDecodingAlgorithm 162
5.4.1.1ViterbiAlgorithm 163
5.5HMMsinDNAandProteinSequences 164
5.6HMMsforModellingSpeechandNoise 165
5.6.1ModellingSpeech 165
5.6.2HMM-BasedEstimationofSignalsinNoise 166
5.6.3SignalandNoiseModelCombinationandDecomposition 167
5.6.4HiddenMarkovModelCombination 168
5.6.5DecompositionofStateSequencesofSignalandNoise 169
5.6.6HMM-BasedWienerFilters 169
5.6.7ModellingNoiseCharacteristics 170
5.7Summary 171
Bibliography 171
6 LeastSquareErrorWiener-KolmogorovFilters 173
6.1LeastSquareErrorEstimation:Wiener-KolmogorovFilter 173
6.1.1DerivationofWienerFilterEquation 174
6.1.2CalculationofAutocorrelationofInputandCross-CorrelationofInputand
DesiredSignals 177
6.2Block-DataFormulationoftheWienerFilter 178
6.2.1QRDecompositionoftheLeastSquareErrorEquation 179
6.3InterpretationofWienerFilterasProjectioninVectorSpace 179
6.4AnalysisoftheLeastMeanSquareErrorSignal 181
6.5FormulationofWienerFiltersintheFrequencyDomain 182
6.6SomeApplicationsofWienerFilters 183
6.6.1WienerFilterforAdditiveNoiseReduction 183
6.6.2WienerFilterandSeparabilityofSignalandNoise 185
6.6.3TheSquare-RootWienerFilter 186
6.6.4WienerChannelEqualiser 187
6.6.5Time-AlignmentofSignalsinMulti-channel/Multi-sensorSystems 187
6.7ImplementationofWienerFilters 188
6.7.1ChoiceofWienerFilterOrder 189
6.7.2ImprovementstoWienerFilters 190
6.8Summary 191
Bibliography 191
7 AdaptiveFilters: Kalman, RLS,LMS 193
7.1Introduction 194
7.2State-SpaceKalmanFilters 195
7.2.1DerivationofKalmanFilterAlgorithm 197
7.2.2RecursiveBayesianFormulationofKalmanFilter 200
7.2.3MarkovianPropertyofKalmanFilter 201
7.2.4ComparisonofKalmanfilterandhiddenMarkovmodel 202
7.2.5ComparisonofKalmanandWienerFilters 202
7.3ExtendedKalmanFilter(EFK) 206
7.4UnscentedKalmanFilter(UFK) 208
7.5SampleAdaptiveFilters–LMS,RLS 211
xii Contents
7.6RecursiveLeastSquare(RLS)AdaptiveFilters 213
7.6.1MatrixInversionLemma 214
7.6.2RecursiveTime-updateofFilterCoefficients 215
7.7TheSteepest-DescentMethod 217
7.7.1ConvergenceRate 219
7.7.2Vector-ValuedAdaptationStepSize 220
7.8LeastMeanSquaredError(LMS)Filter 220
7.8.1LeakyLMSAlgorithm 220
7.8.2NormalisedLMSAlgorithm 221
7.8.2.1DerivationoftheNormalisedLMSAlgorithm 221
7.8.2.2Steady-StateErrorinLMS 222
7.9Summary 223
Bibliography 224
8 LinearPredictionModels 227
8.1LinearPredictionCoding 227
8.1.1Predictability,InformationandBandwidth 228
8.1.2ApplicationsofLPModelinSpeechProcessing 229
8.1.3Time-DomainDescriptionofLPModels 229
8.1.4FrequencyResponseofLPModelanditsPoles 230
8.1.5CalculationofLinearPredictorCoefficients 232
8.1.6EffectofEstimationofCorrelationFunctiononLPModelSolution 233
8.1.7TheInverseFilter:SpectralWhitening,De-correlation 234
8.1.8ThePredictionErrorSignal 235
8.2Forward,BackwardandLatticePredictors 236
8.2.1AugmentedEquationsforForwardandBackwardPredictors 238
8.2.2Levinson–DurbinRecursiveSolution 238
8.2.2.1Levinson–DurbinAlgorithm 240
8.2.3LatticePredictors 240
8.2.4AlternativeFormulationsofLeastSquareErrorPrediction 241
8.2.4.1Burg’sMethod 241
8.2.5SimultaneousMinimisationoftheBackwardandForwardPredictionErrors 242
8.2.6PredictorModelOrderSelection 242
8.3Short-TermandLong-TermPredictors 243
8.4MAPEstimationofPredictorCoefficients 245
8.4.1ProbabilityDensityFunctionofPredictorOutput 245
8.4.2UsingthePriorpdfofthePredictorCoefficients 246
8.5Formant-TrackingLPModels 247
8.6Sub-BandLinearPredictionModel 248
8.7SignalRestorationUsingLinearPredictionModels 249
8.7.1Frequency-DomainSignalRestorationUsingPredictionModels 251
8.7.2ImplementationofSub-BandLinearPredictionWienerFilters 253
8.8Summary 254
Bibliography 254
9 EigenvalueAnalysisandPrincipalComponentAnalysis 257
9.1Introduction–LinearSystemsandEigenAnalysis 257
9.1.1AGeometricInterpretationofEigenvaluesandEigenvectors 258
9.2EigenVectorsandEigenvalues 261
9.2.1MatrixSpectralTheorem 263
9.2.2ComputationofEigenvaluesandEigenVectors 263
Contents xiii
9.3PrincipalComponentAnalysis(PCA) 264
9.3.1ComputationofPCA 265
9.3.2PCAAnalysisofImages:Eigen-ImageRepresentation 265
9.3.3PCAAnalysisofSpeechinWhiteNoise 266
9.4Summary 269
Bibliography 270
10PowerSpectrumAnalysis 271
10.1PowerSpectrumandCorrelation 271
10.2FourierSeries:RepresentationofPeriodicSignals 272
10.2.1ThePropertiesofFourier’sSinusoidalBasisFunctions 272
10.2.2TheBasisFunctionsofFourierSeries 273
10.2.3FourierSeriesCoefficients 274
10.3FourierTransform:RepresentationofNon-periodicSignals 274
10.3.1DiscreteFourierTransform 276
10.3.2Frequency-TimeResolutions:TheUncertaintyPrinciple 277
10.3.3Energy-SpectralDensityandPower-SpectralDensity 278
10.4Non-ParametricPowerSpectrumEstimation 279
10.4.1TheMeanandVarianceofPeriodograms 279
10.4.2AveragingPeriodograms(BartlettMethod) 280
10.4.3WelchMethod:AveragingPeriodogramsfromOverlappedandWindowed
Segments 280
10.4.4Blackman–TukeyMethod 282
10.4.5PowerSpectrumEstimationfromAutocorrelationofOverlapped
Segments 282
10.5Model-BasedPowerSpectrumEstimation 283
10.5.1Maximum–EntropySpectralEstimation 283
10.5.2AutoregressivePowerSpectrumEstimation 285
10.5.3Moving-AveragePowerSpectrumEstimation 286
10.5.4AutoregressiveMoving-AveragePowerSpectrumEstimation 286
10.6High-ResolutionSpectralEstimationBasedonSubspaceEigen-Analysis 287
10.6.1PisarenkoHarmonicDecomposition 287
10.6.2MultipleSignalClassification(MUSIC)SpectralEstimation 289
10.6.3EstimationofSignalParametersviaRotationalInvariance
Techniques(ESPRIT) 291
10.7Summary 293
Bibliography 293
11Interpolation–ReplacementofLostSamples 295
11.1Introduction 295
11.1.1IdealInterpolationofaSampledSignal 296
11.1.2DigitalInterpolationbyaFactorofI 297
11.1.3InterpolationofaSequenceofLostSamples 299
11.1.4TheFactorsThatAffectInterpolationAccuracy 300
11.2PolynomialInterpolation 301
11.2.1LagrangePolynomialInterpolation 302
11.2.2NewtonPolynomialInterpolation 303
11.2.3HermitePolynomialInterpolation 304
11.2.4CubicSplineInterpolation 305
11.3Model-BasedInterpolation 306
11.3.1MaximumAPosterioriInterpolation 307
xiv Contents
11.3.2LeastSquareErrorAutoregressiveInterpolation 308
11.3.3InterpolationBasedonaShort-TermPredictionModel 309
11.3.4InterpolationBasedonLong-TermandShort-termCorrelations 312
11.3.5LSARInterpolationError 314
11.3.6InterpolationinFrequency–TimeDomain 316
11.3.7InterpolationUsingAdaptiveCodeBooks 317
11.3.8InterpolationThroughSignalSubstitution 318
11.3.9LP-HNMModelbasedInterpolation 318
11.4Summary 319
Bibliography 319
12SignalEnhancementviaSpectralAmplitudeEstimation 321
12.1Introduction 321
12.1.1SpectralRepresentationofNoisySignals 322
12.1.2VectorRepresentationofSpectrumofNoisySignals 323
12.2SpectralSubtraction 324
12.2.1PowerSpectrumSubtraction 325
12.2.2MagnitudeSpectrumSubtraction 326
12.2.3SpectralSubtractionFilter:RelationtoWienerFilters 326
12.2.4ProcessingDistortions 327
12.2.5EffectofSpectralSubtractiononSignalDistribution 328
12.2.6ReducingtheNoiseVariance 329
12.2.7FilteringOuttheProcessingDistortions 329
12.2.8Non-LinearSpectralSubtraction 330
12.2.9ImplementationofSpectralSubtraction 332
12.3BayesianMMSESpectralAmplitudeEstimation 333
12.4EstimationofSignaltoNoiseRatios 335
12.5ApplicationtoSpeechRestorationandRecognition 336
12.6Summary 338
Bibliography 338
13ImpulsiveNoise:Modelling,DetectionandRemoval 341
13.1ImpulsiveNoise 341
13.1.1DefinitionofaTheoreticalImpulseFunction 341
13.1.2TheShapeofaRealImpulseinaCommunicationSystem 342
13.1.3TheResponseofaCommunicationSystemtoanImpulse 343
13.1.4TheChoiceofTimeorFrequencyDomainforProcessingofSignals
DegradedbyImpulsiveNoise 343
13.2AutocorrelationandPowerSpectrumofImpulsiveNoise 344
13.3ProbabilityModelsofImpulsiveNoise 345
13.3.1Bernoulli–GaussianModelofImpulsiveNoise 346
13.3.2Poisson–GaussianModelofImpulsiveNoise 346
13.3.3ABinary-StateModelofImpulsiveNoise 347
13.3.4HiddenMarkovModelofImpulsiveandBurstNoise 348
13.4ImpulsiveNoiseContamination,SignaltoImpulsiveNoiseRatio 349
13.5MedianFiltersforRemovalofImpulsiveNoise 350
13.6ImpulsiveNoiseRemovalUsingLinearPredictionModels 351
13.6.1ImpulsiveNoiseDetection 352
13.6.2AnalysisofImprovementinNoiseDetectability 353
13.6.3Two-SidedPredictorforImpulsiveNoiseDetection 355
13.6.4InterpolationofDiscardedSamples 355
Contents xv
13.7RobustParameterEstimation 355
13.8RestorationofArchivedGramophoneRecords 357
13.9Summary 358
Bibliography 358
14TransientNoisePulses 359
14.1TransientNoiseWaveforms 359
14.2TransientNoisePulseModels 361
14.2.1NoisePulseTemplates 361
14.2.2AutoregressiveModelofTransientNoisePulses 362
14.2.3HiddenMarkovModelofaNoisePulseProcess 363
14.3DetectionofNoisePulses 364
14.3.1MatchedFilterforNoisePulseDetection 364
14.3.2NoiseDetectionBasedonInverseFiltering 365
14.3.3NoiseDetectionBasedonHMM 365
14.4RemovalofNoisePulseDistortions 366
14.4.1AdaptiveSubtractionofNoisePulses 366
14.4.2AR-basedRestorationofSignalsDistortedbyNoisePulses 367
14.5Summary 369
Bibliography 369
15EchoCancellation 371
15.1Introduction:AcousticandHybridEcho 371
15.2EchoReturnTime:TheSourcesofDelayinCommunicationNetworks 373
15.2.1Transmissionlink(electromagneticwavepropagation)delay 374
15.2.2Speechcoding/decodingdelay 374
15.2.3Networkprocessingdelay 374
15.2.4De-Jitterdelay 375
15.2.5Acousticechodelay 375
15.3TelephoneLineHybridEcho 375
15.3.1EchoReturnLoss 376
15.4Hybrid(TelephoneLine)EchoSuppression 377
15.5AdaptiveEchoCancellation 377
15.5.1EchoCancellerAdaptationMethods 379
15.5.2ConvergenceofLineEchoCanceller 380
15.5.3EchoCancellationforDigitalDataTransmission 380
15.6AcousticEcho 381
15.7Sub-BandAcousticEchoCancellation 384
15.8EchoCancellationwithLinearPredictionPre-whitening 385
15.9Multi-InputMulti-OutputEchoCancellation 386
15.9.1StereophonicEchoCancellationSystems 386
15.9.2Non-uniquenessProbleminMIMOEchoChannelIdentification 387
15.9.3MIMOIn-CabinCommunicationSystems 388
15.10Summary 389
Bibliography 389
16ChannelEqualisationandBlindDeconvolution 391
16.1Introduction 391
16.1.1TheIdealInverseChannelFilter 392
16.1.2EqualisationError,ConvolutionalNoise 393
16.1.3BlindEqualisation 394
xvi Contents
16.1.4Minimum-andMaximum-PhaseChannels 396
16.1.5WienerEqualiser 396
16.2BlindEqualisationUsingChannelInputPowerSpectrum 398
16.2.1HomomorphicEqualisation 398
16.2.2HomomorphicEqualisationUsingaBankofHigh-PassFilters 400
16.3EqualisationBasedonLinearPredictionModels 400
16.3.1BlindEqualisationThroughModelFactorisation 401
16.4BayesianBlindDeconvolutionandEqualisation 402
16.4.1ConditionalMeanChannelEstimation 403
16.4.2Maximum-LikelihoodChannelEstimation 403
16.4.3MaximumAPosterioriChannelEstimation 404
16.4.4ChannelEqualisationBasedonHiddenMarkovModels 404
16.4.5MAPChannelEstimateBasedonHMMs 406
16.4.6ImplementationsofHMM-BasedDeconvolution 407
16.5BlindEqualisationforDigitalCommunicationChannels 409
16.5.1LMSBlindEqualisation 410
16.5.2EqualisationofaBinaryDigitalChannel 413
16.6EqualisationBasedonHigher-OrderStatistics 414
16.6.1Higher-OrderMoments,CumulantsandSpectra 414
16.6.1.1Cumulants 415
16.6.1.2Higher-OrderSpectra 416
16.6.2Higher-OrderSpectraofLinearTime-InvariantSystems 416
16.6.3BlindEqualisationBasedonHigher-OrderCepstra 417
16.6.3.1Bi-Cepstrum 418
16.6.3.2Tri-Cepstrum 419
16.6.3.3CalculationofEqualiserCoefficientsfromtheTri-cepstrum420
16.7Summary 420
Bibliography 421
17SpeechEnhancement:NoiseReduction,BandwidthExtension
andPacketReplacement 423
17.1AnOverviewofSpeechEnhancementinNoise 424
17.2Single-InputSpeechEnhancementMethods 425
17.2.1ElementsofSingle-InputSpeechEnhancement 425
17.2.1.1SegmentationandWindowingofSpeechSignals 426
17.2.1.2SpectralRepresentationofSpeechandNoise 426
17.2.1.3LinearPredictionModelRepresentationofSpeechandNoise426
17.2.1.4Inter-FrameandIntra-FrameCorrelations 427
17.2.1.5SpeechEstimationModule 427
17.2.1.6ProbabilityModelsofSpeechandNoise 427
17.2.1.7CostofErrorFunctionsinSpeechEstimation 428
17.2.2WienerFilterforDe-noisingSpeech 428
17.2.2.1WienerFilterBasedonLinearPredictionModels 429
17.2.2.2HMM-BasedWienerFilters 429
17.2.3SpectralSubtractionofNoise 430
17.2.3.1SpectralSubtractionUsingLPModelFrequency
Response 431
17.2.4BayesianMMSESpeechEnhancement 432
17.2.5KalmanFilterforSpeechEnhancement 432
17.2.5.1KalmanState-SpaceEquationsofSignalandNoiseModels433
Contents xvii
17.2.6SpeechEnhancementUsingLP-HNMModel 435
17.2.6.1OverviewofLP-HNMEnhancementSystem 436
17.2.6.2FormantEstimationfromNoisySpeech 437
17.2.6.3Initial-CleaningofNoisySpeech 437
17.2.6.4FormantTracking 437
17.2.6.5HarmonicPlusNoiseModel(HNM)ofSpeechExcitation 438
17.2.6.6FundamentalFrequencyEstimation 439
17.2.6.7EstimationofAmplitudesHarmonicsofHNM 439
17.2.6.8EstimationofNoiseComponentofHNM 440
17.2.6.9KalmanSmoothingofTrajectoriesofFormantsandHarmonics440
17.3SpeechBandwidthExtension–SpectralExtrapolation 442
17.3.1LP-HNMModelofSpeech 443
17.3.2ExtrapolationofSpectralEnvelopeofLPModel 444
17.3.2.1PhaseEstimation 445
17.3.2.2CodebookMappingoftheGain 445
17.3.3ExtrapolationofSpectrumofExcitationofLPModel 446
17.3.3.1SensitivitytoPitch 446
17.4InterpolationofLostSpeechSegments–PacketLossConcealment 447
17.4.1PhasePrediction 450
17.4.2CodebookMapping 452
17.4.2.1EvaluationofLP-HNMInterpolation 453
17.5Multi-InputSpeechEnhancementMethods 455
17.5.1Beam-formingwithMicrophoneArrays 457
17.5.1.1SpatialConfigurationofArrayandTheDirectionofReception458
17.5.1.2DirectionalofArrival(DoA)andTimeofArrival(ToA) 459
17.5.1.3SteeringtheArrayDirection:EqualisationoftheToAs
attheSensors 459
17.5.1.4TheFrequencyResponseofaDelay-SumBeamformer 460
17.6SpeechDistortionMeasurements 462
17.6.1Signal-to-NoiseRatio–SNR 462
17.6.2SegmentalSignaltoNoiseRatio– SNR
462
seg
17.6.3Itakura–SaitoDistance–ISD 463
17.6.4HarmonicityDistance–HD 463
17.6.5DiagnosticRhymeTest–DRT 463
17.6.6MeanOpinionScore–MOS 464
17.6.7PerceptualEvaluationofSpeechQuality–PESQ 464
Bibliography 464
18Multiple-InputMultiple-OutputSystems,IndependentComponentAnalysis 467
18.1Introduction 467
18.2Anoteoncomparisonofbeam-formingarraysandICA 469
18.3MIMOSignalPropagationandMixingModels 469
18.3.1InstantaneousMixingModels 469
18.3.2Anechoic,DelayandAttenuation,MixingModels 470
18.3.3ConvolutionalMixingModels 471
18.4IndependentComponentAnalysis 472
18.4.1ANoteonOrthogonal,OrthonormalandIndependent 473
18.4.2StatementofICAProblem 474
18.4.3BasicAssumptionsinIndependentComponentAnalysis 475
18.4.4TheLimitationsofIndependentComponentAnalysis 475
xviii Contents
18.4.5WhyamixtureoftwoGaussiansignalscannotbeseparated? 476
18.4.6TheDifferenceBetweenIndependentandUncorrelated 476
18.4.7IndependenceMeasures;EntropyandMutualInformation 477
18.4.7.1DifferentialEntropy 477
18.4.7.2MaximumValueofDifferentialEntropy 477
18.4.7.3MutualInformation 478
18.4.7.4TheEffectofaLinearTransformationonMutualInformation479
18.4.7.5Non-GaussianityasaMeasureofIndependence 480
18.4.7.6Negentropy:AmeasureofNon-GaussianityandIndependence480
18.4.7.7FourthOrderMoments–Kurtosis 481
18.4.7.8Kurtosis-basedContrastFunctions–Approximationsto
EntropicContrast 481
18.4.8Super-GaussianandSub-GaussianDistributions 482
18.4.9Fast-ICAMethods 482
18.4.9.1Gradientsearchoptimisationmethod 483
18.4.9.2Newtonoptimisationmethod 483
18.4.10Fixed-pointFastICA 483
18.4.11ContrastFunctionsandInfluenceFunctions 484
18.4.12ICABasedonKurtosisMaximization–ProjectionPursuitGradientAscent 485
18.4.13JadeAlgorithm–IterativeDiagonalisationofCumulantMatrices 487
18.5Summary 490
Bibliography 490
19SignalProcessinginMobileCommunication 491
19.1IntroductiontoCellularCommunication 491
19.1.1ABriefHistoryofRadioCommunication 492
19.1.2CellularMobilePhoneConcept 493
19.1.3OutlineofaCellularCommunicationSystem 494
19.2CommunicationSignalProcessinginMobileSystems 497
19.3Capacity,Noise,andSpectralEfficiency 498
19.3.1SpectralEfficiencyinMobileCommunicationSystems 500
19.4Multi-pathandFadinginMobileCommunication 500
19.4.1Multi-pathPropagationofElectromagneticSignals 501
19.4.2RakeReceiversforMulti-pathSignals 502
19.4.3SignalFadinginMobileCommunicationSystems 502
19.4.4Large-ScaleSignalFading 504
19.4.5Small-ScaleFastSignalFading 504
19.5SmartAntennas–Space–TimeSignalProcessing 505
19.5.1SwitchedandAdaptiveSmartAntennas 506
19.5.2Space–TimeSignalProcessing–DiversitySchemes 506
19.6Summary 508
Bibliography 508

oasis87

多谢分享！！

yhydfbb

我有中文版的，现在可以对照起来看，好锻炼一下英文能力

apteye

xiexie

apteye

xiexie

lokta

这个牛X

forestmeng

好东西,大家加油,不要沉了,自己顶一下!!!

fpga_sh

谢谢楼主提供资料

fpga_sh

谢谢楼主提供资料

amin2008

看看先