COMP30291: Digital Media Processing (2009-2010)
This course is concerned with the application of digital signal processing (DSP) to speech, music and video. Students will gain an appreciation of the software, the design tools currently available and some of the most important applications of media processing in general purpose and embedded systems.
Understanding of the significance of digital signal processing in multi-media processing, storage and communications.
Familiarity with fundamental concepts such as 'linearity' , 'time-invariance', 'impulse response', 'convolution', 'frequency- response', 'z-transforms' and the 'discrete time Fourier transform' as applied to signal processing systems.
A knowledge of digital filters and their application to digitised sound and images.
Understanding of how 'FIR type' and 'IIR type' digital filters may be designed and implemented in software.
Ability to use the 'MATLAB' language and 'signal processing toolboxes' for designing, implementing and simulating digital signal processing (DSP) operations as applied to speech, music and images.
Ability to specify the 'real time' implementation of DSP operations using special purpose fixed point 'DSP microprocessors'.
Understanding of the effects of analogue/digital conversion.
Introduction to the discrete Fourier transform (DFT), its applications and its implementation by FFT techniques.
Knowledge of how 'MP3' music compresssion is achieved using the FFT and perceptual masking.
Assessment of Learning outcomes2 hour written examination.
Contribution to Programme Learning OutcomesA2, A3, A5, B1, B2, B3, C4.
Section 1: Introduction (1 lecture)
Definition of continuous time (analogue), discrete time and digital signals. Sampling and quantisation in general terms.
Introduction to signal processing as applied to speech, music and multimedia.
Section 2: Analogue and digital filters (2 lectures)
Transfer function and frequency-response of a filter.
Low-pass, band-pass & band-stop filters.
Butterworth low-pass gain response approximation.
Section 3: Digital signal processing (4 lectures)
Recursive and non-recursive difference equations. Signal flow-graphs and their implementation by simple computer programs.
Linearity, time invariance and impulse-response for discrete time systems.
Definition of finite impulse response (FIR) and infinite impulse response (IIR) type digital filters. Stability and causality. Time-domain convolution. Frequency response as discrete time Fourier transform (DTFT) of impulse-response. Gain and phase responses. Linear phase and group delay. Inverse DTFT.
Use of MATLAB for analysing digital filters.
Section 4: FIR type digital filters (2 lectures)
Design of FIR digital filters by the Fourier series approximation method.
Implementation on personal computers and in real time on dsp chips.
Effect of increasing order and use of non-rectangular windows. Alternative methods.
Section 5: IIR type digital filters and the z-transform (4 lectures)
System function, H(z), as z-transform of impulse response.
Relationship between system function, difference equation, signal flow-graph and software implementation of FIR and IIR type digital filters.
Poles and zeros. Distance rule for estimating the gain response of a digital filter from an Argand diagram (z-plane) of poles and zeros. Design of a digital IIR "notch" filter and a resononator by pole/zero placement. Application to these filters to sound recordings. Design of IIR type digital filters using MATLAB. (3 lectures)
Section 6: Digital processing of multimedia signals (3 lectures)
Sampling theory, aliasing and the effect of quantisation and 'sample and hold' reconstruction. Sampling rate conversion and oversampling to simplify analogue filters. Overall design of digital systems for processing speech, music and multimedia.
Section 7: The discrete Fourier transform & its applications (3 lectures)
Derivation of DFT from DTFT. Inverse DFT. Effects of windowing and frequency- domain sampling. Non-rectangular windows. The 'fast Fourier transform' algorithm (FFT). Use of the FFT for spectral estimation and media processing. 2-D FFT, Case studies in MATLAB.
Section 8: Processing speech, music & video (3 lectures)
Digitising speech with bit-rate compression (A-law, ADPCM, CELP). Digitising music (CDs, MPEG & MP3)and video (JPEG & MPEG)
Discussion of exercises and problems.
S.W. Smith, 'Scientist and Engineer's Guide to Digital Signal Processing' is available complete at: http://www.dspguide.com/
Lecture notes, supporting material & solutions to examinations & other problems available at:
Core TextTitle: Digital signal processing: a practical guide for engineers and scientists
Author: Smith, Steven W.