Ceska verze teto stranky

BE2M31DSP - Advanced DSP methods
Seminar schedule - summer semester 2022/2023


SEMINAR SCHEDULE:

  1. Tu - 21.2.2023 - Selected characteristics of discrete-time signals and systems, signal modelling
    - Reminder of knowledge from BE2M31DSPA
    - Modelling of signals using AR, MA, and ARMA models
    Seminar guidelines

  2. Tu 28.2.2023 - Basis of LPC
    - LPC analysis, autocorrelation method (Yule-Walker equations)
    - LPC spectrum (smoothed estimation of PSD)
    Seminar guidelines

  3. Tu - 7.3.2023 - Methods for delay estimation
    - Delay estimation based on cross-correlation
    - Delay estimation based on cross power spectral density (CPSD)
    - Estimation dsicrete-system transfer function
    Seminar guidelines

  4. Tu - 14.3.2023 - Applications of coherence analysis
    - Definition of coherence function and estimation methods
    - cohergram
    - Time dependancy of average coherence
    - Coherence applications (noise reduction and delay detection)
    Seminar guidelines

  5. Tu - 21.3.2023 - Cepstral analysis - basic properties and applications
    - Definition, properties, and basic applications of real and complex cepstrum
    - Liftering and homomorphic deconvolution
    Seminar guidelines

  6. Tu - 28.3.2023 - TEST 1 - Cepstral distance and its applications
    - computation of DFT and LPC cepstrum, computation of cepstral distance
    - measurement of signal distortion
    - detection of frame boundaries in non-stationary signal
    Seminar guidelines

  7. Tu - 4.4.2023 - Discrete cosine transform
    - Definitions of DCT and computation approaches, relationship to DFT
    - Using DCT for real cepstrum computation
    - Compression using DCT
    Seminar guidelines

  8. Tu - 11.4.2023 - Frequency-domain suppression of stationary additive noise
    - suppression of stationary noise in a stationary signal
    - Wiener filtering, DCT-based Wiener filtering, spectral subtraction
    Seminar guidelines

  9. Tu - 18.4.2023 - Frequency-domain suppression of stationary additive noise in non-stationary signal
    - adaptive variants of algorithms basedon Wiener filtering, DCT, and spectral subtraction,
    Seminar guidelines

  10. Tu - 25.4.2023 - Inverse filtering and Wiener filtering with a distortion model
    - compensation of distortion given by trasmitting channel using inverse fitering
    - inverse filtering in noisy environment
    - Wiener-filter-based suppression of noise influence during inverse filtering
    Seminar guidelines

  11. Tu - 2.5.2023 - Pricipal Component analysis and KLT transform
    - Eigenvalues and eigenvectors of signal covariance matrix
    - Karhunen-Loeve transform (KLT)
    - Comparison of KLT, DCT, and DFT
    Seminar guidelines

  12. Tu - 9.5.2023 - Seminar is cancelled (CHANGE OF SCHEDULING)

  13. Tu - 16.5.2023 - TEST 2 - ICA-based blid separation of signals
    - Multichannel linear mixture of signals
    - Mixing and inverse matrices
    - Blind separation, FastICA
    Seminar guidelines

  14. Tu - 23.5.2023 - Wavelet transform (WT), filter banks (QMFB), Noise suppression based on WT
    HOMEWORK
    - Decomposition of signal into more bands using quadrature-mirror filters (QMF)
    COMPLETION OF THE TASK AT SEMINAR
    - Noise suppression using WT (QMFB)
    Seminar guidelines