This repository contains the complete set of exercises and supplementary materials for the Digital Speech Signal Processing course conducted during the Summer Semester 2020 at the University of Paderborn as part of the M.Sc. program.
This course covers fundamental and advanced topics in digital speech signal processing, progressing from basic signal analysis to sophisticated audio processing techniques.
Ex01: Continuous and Discrete Time Fourier Transform
- Signal representation in time and frequency domains
- Fourier analysis fundamentals
Ex02: Sampling and Reconstruction
- Nyquist sampling theorem
- Signal reconstruction from samples
- Quantization effects
- Aliasing phenomena
Ex03: Energy, Discretization and Parseval's Theorem
- Signal energy and power calculations
- Parseval's theorem applications
- Discretization effects
Ex04: DFT and Computational Effort
- Discrete Fourier Transform implementation
- Computational complexity analysis
- FFT algorithm efficiency
Ex05: Partitioned Block Overlap Save
- Convolution techniques for long signals
- Block processing methods
- Overlap-save algorithm implementation
Ex06: Window Functions and Leakage Effect
- Spectral leakage analysis
- Window function characteristics
- Frequency resolution trade-offs
Ex07: Periodograms and ACF Estimation and MFCC
- Power spectral density estimation
- Autocorrelation function analysis
- Mel-frequency cepstral coefficients
- Feature extraction for speech processing
Ex08: Linear Estimator for Auto-regressive Processes (with LPC)
- Linear prediction coding
- Auto-regressive modeling
- Speech analysis applications
Ex09: Noise Reduction
- Spectral subtraction methods
- Wiener filtering
- Speech enhancement techniques
Ex11: Adaptive Filters
- LMS and NLMS algorithms
- Adaptive noise cancellation
- System identification applications
Supplement 1: Fourier Series
- Fourier series decomposition
- Harmonic analysis
- Signal synthesis
Supplement 2: From Fourier Series to FFT
- Algorithm evolution
- Computational optimization
- FFT implementation details
Supplement 3: Short Term Fourier Transform
- Time-frequency analysis
- Spectrogram representation
- STFT windowing strategies
The exercises follow a logical progression building upon fundamental concepts:
Ex01 (Fourier Transform) → Ex02 (Sampling) → Ex03 (Energy & Parseval) → Ex04 (DFT/FFT)
↓
Ex06 (Window Functions) → Ex05 (Block Convolution)
↓
Ex07 (Spectral Analysis) → Ex08 (Linear Prediction)
↓
Ex09 (Noise Reduction) → Ex11 (Adaptive Filters)
Supplementary materials provide theoretical foundations:
Supplement 1 (Fourier Series) → Supplement 2 (FFT Development) → Supplement 3 (STFT)
The exercises demonstrate various signal processing techniques using:
- Python implementations with NumPy and SciPy
- MATLAB scripts for algorithm development
- Jupyter notebooks for interactive analysis
- Real audio signal examples
- Institution: University of Paderborn
- Program: M.Sc. (Master of Science)
- Semester: Summer Semester 2020 (April - September)
- Duration: 6 months of progressive learning
Each exercise directory contains:
- Template files for initial implementation
- Solution files with complete implementations
- Scratch files for development and experimentation
- Supporting audio files for real-world testing
- Comprehensive documentation and analysis
Basic understanding of:
- Signal and systems theory
- Linear algebra
- Probability theory
- Programming fundamentals (Python/MATLAB)
Upon completion, students will have practical experience with:
- Digital signal processing fundamentals
- Speech analysis and synthesis techniques
- Audio enhancement and noise reduction
- Adaptive filtering applications
- Feature extraction for speech processing