Applications of DSP to Audio and Acoustics
Applications of Digital Signal Processing to Audio and Acoustics
Mark Kahrs and Karlheinz Brandenburg, editors
Kluwer Academic Publishers
March 1998
ISBN 0-7923-8130-0
$124.00 (sorry!)
To order, see publisher's page
Audio quality determination based on perceptual measurement techniques
John Beerends
KPN Research
- Introduction
- Basic measuring philosophy
- Subjective versus objective perceptual testing
- Psychoacoustic fundamentals of calculating the
- Computation of the internal sound representation
- The perceptual audio quality measure (PAQM)
- Validation of the PAQM on speech and music codec databases
- Cognitive effects in judging audio quality
- ITU Standardization
- ITU-T, speech quality
- ITU-R, audio quality
- Conclusions
Perceptual Coding of High Quality Digital Audio
Karlheinz Brandenburg
FhG-IIS
- Introduction
- Application considerations
- Application areas of audio coding
- Requirements for audio coding systems
- Hearing
- Redundancy
- Source coding versus perceptual coding
- Some Facts about Psychoacoustics
- Masking in the Frequency Domain
- Masking in the Time Domain
- Variability between listeners
- Basic ideas of perceptual coding
- Basic block diagram
- Additional coding tools
- Perceptual Entropy
- Description of coding tools
- Filter banks
- QMF filter banks
- Wavelet based filter banks
- Polyphase filter banks
- Time domain aliasing cancellation based filter banks
- Hybrid filter banks
- Alias reduction for hybrid filter banks
- Adaptive block switching
- Adaptive filter banks
- Perceptual models
- A trivial example
- Estimation of tonality
- MPEG-1 perceptual model 2
- Quantization and coding
- Block companding
- Non-uniform scalar quantization
- Vector quantization
- Noise allocation followed by scalar quantization and
- Huffman coding
- Short time buffering
- Joint stereo coding
- Pitfalls of stereo coding
- General ideas
- M/S stereo coding
- Intensity stereo coding
- Coupling channels
- Prediction
- Pointers to early systems
- MPEG Audio
- MPEG-1 Layer 1 and Layer 2
- Quantization and Coding
- Layer 2 Quantization and Coding
- Bit allocation
- MPEG-1 Layer 3
- MPEG-2 Audio
- Backwards compatible multichannel coding
- Coding at lower sampling frequencies
- MPEG-2 Advanced Audio Coding (MPEG-2 AAC)
- Gain Control
- Filterbank
- Temporal Noise Shaping (TNS)
- Intensity Coding/Coupling
- Prediction
- M/S Stereo Coding
- Scalefactors
- Quantization
- Noiseless Coding
- Perceptual Model
- MPEG-4 Audio
- The focus of MPEG-4
- Bridging the gap between signal synthesis, speech coding and ...
- Scaleable audio coding
- Current Research Topics
- Filterbanks
- Perceptual Models
- Quantization and Coding
- Lossless and near lossless coding
- Conclusions
Reverberation Algorithms
Bill Gardner
MIT Media Lab
- Introduction
- Reverberation as a linear filter
- Approaches to reverberation algorithms
- The physical approach
- The perceptual approach
- Physical and Perceptual Background
- Measurement of reverberation
- Early reverberation
- Perceptual effects of early echoes
- Reverberation time
- Modal description of reverberation
- Statistical model for reverberation
- Subjective and objective measures of late reverberation
- Summary of framework
- Modeling Early Reverberation
- Comb and Allpass Reverberators
- Schroeder's reverberator
- The parallel comb filter
- Modal density and echo density
- Producing uncorrelated outputs
- Moorer's reverberator
- Allpass reverberators
- Feedback Delay Networks
- Jot's reverberator
- Unitary feedback loops
- Absorptive delays
- Waveguide reverberators
- Lossless prototype structures
- Implementation of absorptive and correction filters
- Multirate algorithms
- Time-varying algorithms
- Conclusions
Digital Audio Restoration
Simon Godsill and Peter Rayner (Cambridge)
Olivier Cappe (ENST)
- Introduction
- Modelling of audio signals
- Click Removal
- Modelling of clicks
- Detection
- Autoregressive (AR) model-based Click Detec
- Replacement of corrupted samples
- Autoregressive interpolation
- Other methods
- Statistical methods for the treatment of clicks
- Correlated Noise Pulse Removal
- Background noise reduction
- Background noise reduction by short-time spectral attenuation
- Discussion
- Template-based methods
- Model-based separation methods
- Summary
- Pitch variation defects
- Reduction of Non-linear Amplitude Distortion
- Distortion Modelling
- Non-linear Signal Models
- The Volterra Series
- NARMA Modelling
- Application of Non-linear models to Distortion Reduction
- Memoryless Non-linearity
- Non-linearity with Memory
- Parameter Estimation
- Examples
- Discussion
- General overview
- Suppression rules
- Evaluation
- The musical noise phenomenon
- Current trends and perspectives
- Frequency domain estimation
- Other areas
- Conclusion and Future Trends
Digital Audio System Architecture
Mark Kahrs
CAIP Center, Rutgers University
- Introduction
- Input/Output
- Analog/Digital Conversion
- Fixed Point Converters
- Floating Point Converters
- Sampling clocks
- Processing
- Requirements
- Analysis/Synthesis
- Analysis
- Frequency Analysis
- Processing
- Synthesis
- Linear Synthesis
- Subtractive Synthesis
- Physical Modelling
- Processors
- Sequential (MSI scale) machines
- Sequential single processors
- Custom chips
- Parallel machines
- Conclusion
Signal Processing for Hearing Aids
Jim Kates
Audiologic
- Introduction
- Hearing and Hearing Loss
- Inner Ear
- Retrocochlear and Central Losses
- Summary
- Linear Amplification
- System Description
- Dynamic Range
- Distortion
- Bandwidth
- Feedback Cancellation
- Compression Amplification
- Single-Channel Compression
- Two-Channel Compression
- Multi-Channel Compression
- Single-Microphone Noise Suppression
- Adaptive Analog Filters
- Spectral Subtraction
- Spectral Enhancement
- Multi-Microphone Noise Suppression
- Directional Microphone Elements
- Two-Microphone Adaptive Noise Cancellation
- Arrays with Time-Invariant Weights
- Two-Microphone Adaptive Arrays
- Multi-Microphone Adaptive Arrays
- Performance Comparison in a Real Room
- Cochlear Implants
- Conclusions
Time and Pitch scale modification of audio signals
Jean Laroche
E-mu
- Introduction
- Notations and definitions
- An underlying sinusoidal model for signals
- A definition of time-scale and pitch-scale modification
- Frequency-domain techniques
- Methods based on the short-time Fourier transform
- Definitions and assumptions
- Short-time Fourier transform of a sinusoidal signal
- Time scaling
- Pitch-scaling
- Choice of the analysis parameters
- Methods based on a signal model
- Time-domain techniques
- Principle
- Pitch independent methods
- The analog origin
- Digital counterpart
- Input-time/Output-time characteristic
- The problem of tempo
- Periodicity-driven methods
- Methods based on waveform similarity
- The PSOLA method
- Formant modification
- Time-domain techniques
- Frequency-domain techniques
- Discussion
- Generic problems associated with time or pitch scaling
- Reverberation and shape invariance
- Transient smearing
- Time-domain vs frequency-domain techniques
Wavetable Sampling Synthesis
Dana Massie
E-mu
- Background and Introduction
- Transition to Digital
- Flourishing of Digital Synthesis Methods
- Sampling vs. Synthesis
- Wavetable Sampling Synthesis
- Playback of digitized musical instrument events
- Entire note - not single period
- Effect of storing attacks
- Loops
- Pitch Shifting Technologies
- Asynchronous Pitch Shifting
- Synchronous Pitch Shifting
- Sample rate conversion
- Frequency Scaling
- Formant Re-scaling
- Sample Rate Conversion Techniques
- Linear Interpolation Table Lookup Oscillators
- How Many Fractional Phase Register Bits are Needed
- Looping of sustain
- Backwards-Forwards loops
- Loop body
- Crossfade looping
- Backwards Forwards Crossfade Looping
- Relation of Crossfade Looping to Time and Pitch Scaling
- Perception of periodicity in loops
- Multi-sampling
- Enveloping
- Filtering
- Amplitude variations as a function of velocity
- Mixing or summation of channels
- Multiplexed wavetables
- Conclusion
Sinusoidal Analysis/Synthesis in Audio Processing
Tom Quatieri and Bob McAulay
MIT Lincoln Labs
- Introduction
- Filter Bank Analysis/Synthesis
- Additive Synthesis
- Phase Vocoder
- Motivation for a Sine-Wave Analysis/Synthesis
- Sinusoidal-Based Analysis/Synthesis
- Model
- Estimation of Model Parameters
- Frame-to-Frame Peak Matching
- Synthesis
- Experimental Results
- Applications of the Baseline System
- Time-Scale Modification
- Sound Splicing, Interpolation, and Extrapolation
- Tracking Vibrato
- Time-Frequency Resolution
- Source/Filter Phase Model
- Model
- Phase Coherence in Signal Modification
- Time-Scale Modification
- Dynamic Range Compression
- Revisiting the Filter Bank-Based Approach
- Temporal Envelope
- Instantaneous Invariance
- Short-Time Processing
- Additive Deterministic/Stochastic Model
- Model
- Analysis/Synthesis
- Applications
- Separation of Sound Components
- Signal Modification
- Signal Separation Using a Two-Voice Model
- Formulation of the Separation Problem
- Analysis and Separation
- The Ambiguity Problem
- Pitch and Voicing Estimation
- FM Synthesis
- Principles
- Basic Model
- Generalizations
- Representation of Musical Sound
- Parameter Estimation
- Extensions
- Conclusions
Principles of Digital Waveguide Models of Musical Instruments
Julius Smith
CCRMA, Stanford University
- Introduction
- Antecedents in Speech Modeling
- Physical Models in Music Synthesis
- Summary
- The Ideal Vibrating String
- The Finite Difference Approximation
- General Properties of the FDA
- FDA of the Ideal String
- Traveling-Wave Solution
- Sampling the Traveling Waves
- Relation to Finite Difference Recursion
- Alternative Wave Variables
- Spatial Derivatives
- Force Waves
- Power Waves
- Energy Density Waves
- Root-Power Waves
- Scattering at an Impedance Discontinuity
- The Kelly-Lochbaum and One-Multiply Scattering Junctions
- Normalized Scattering Junctions
- Transformer Normalization
- The Three-Multiply Normalized Scattering Junction
- Junction Passivity
- Passive Kelly-Lochbaum and One-Multiply Junctions
- Passive Four-Multiply Normalized Junctions
- Passive Three-Multiply Normalized Junctions
- Scattering at a Loaded Junction of N Waveguides
- The Lossy One-Dimensional Wave Equation
- Loss Consolidation
- Frequency-Dependent Losses
- The Dispersive One-Dimensional Wave Equation
- Single-Reed Instruments
- Clarinet Overview
- Single-Reed Theory
- Scattering-Theoretic Formulation of the Reed
- Computational Methods
- Practical Details
- Bowed Strings
- Violin Overview
- The Bow-String Scattering Junction
- Conclusions
Created on Feb 5 at 11:18
Mark Kahrs,
CAIP Center,
P.O. Box 1390,
Piscataway, NJ,
08855-1390
Phone: +1 732-445-6348, Fax: +1 732-445-4775
e-mail: kahrs@caip.rutgers.edu
home page:http://www.caip.rutgers.edu/~kahrs