Ceska verze teto stranky

Guidelines to seminar:

• Estimation of delay for 2-channel signals using Cross-Correlation Function (CCF)
  • Determine the delay between channels for the following available signals using CCF computed within short quasi-stationary segments.
  • Realize the computation of CCF only for the case of higly correlated short-time frames.
  • Take into account the following signal pairs
    (all signals mentioned below are in binary format: 16-bit PCM, raw data without header, sampling frequency is 16 kHz, to load them into MATLAB use function loadbin.m)
    • Simulation 1: signal in the channel No.1: SA001S01.CS0 - signal in the second channel No.2 should be created by delaying of signal in channel No. 1 20 samples (IDEAL REFLECTION).
    • Simulation 2: create the signal in the channel 2 by the filtering of the signal in channel 1 using wide-band FIR filter (exactly wide-band high-pass filter - cutoff frequency should be 0.05*fs/2, order 30). Add also small level white noise ot both channels (standard deviation of generated white Gaussain noise should be approx. 0.1-0.01). Do not forget to generate two different (independent) realizations of white noise for both channels.
    • Real signals I: channel 0 (CS0) - SA001S01.CS0, channel 1 (CS1) - SA001S01.CS1, channel 2 (CS2) - SA001S01.CS2, channel 3 (CS3) - SA001S01.CS3,
    • Result: Observe signals for further analysis according to Simulation 1, Simulation 2 same as Real signals I for signals pairs CS0-CS1, CS0-CS2 and CS0-CS3 (always chosen 2 channels into the same plot with different color. Observe estimated delay and varying level of similarity (correlation) between given signals.

    Repeat it for the following other signals in the free time:

    • Real signals II: channel 0 (CS0) - SA176S01.CS0, channel 1 (CS1) - SA176S01.CS1, channel 2 (CS2) - SA176S01.CS2, channel 3 (CS3) - SA176S01.CS3,
    • Real signals III: channel 0 (CS0) - SA005S01.CS0, channel 1 (CS1) - SA005S01.CS1, channel 2 (CS2) - SA005S01.CS2, channel 3 (CS3) - SA005S01.CS3,

• Estimation of delay for 2-channel signals using Cross Power Spectral Density (CPSD)
  • Explain the principle of the computation of Cross Power Spectral Density for two-channel signals and discuss imortant properties of CPSD.
  • Determine the delay between 2 channels for given analyzed speech signals from linear parts of phase of CPSD.
  • Compute it for signal pairs from previous task.
  • Result: CPSD estimation for chosen highly correlated signal frame selected from signals of Simulation 1, Simulation 2 and Real signals I.

  ---

• Estimation of impulse response of given room and modelling of convolution distortion
  • Determine transfer function (frequency response) and transfer function in a room with acoustic reverberations on the basis of signals x1, x2, and x3 collected in three channels (using microphones of different quality) and the signal x0 collected by high-quality head-set microphone:
    x0 - SA001S04.CS0, x1 - SA001S04.CS1, x2 - SA001S04.CS2, x3 - SA001S04.CS3
    (all signals mentioned below are in binary format: 16-bit PCM, raw data without header, sampling frequency is 16 kHz, to load them into MATLAB use function loadbin.m)
  • Transfer function (frequency response) between signals x0 and x1 (or x2, x3) should be determinee using available function tfestimate.
  • Estimate impulse response on the basis of estimated transfer function (frequency response) which can be used then for a modelling of reverberations in given room.
  • Result: Observe :
    • estimation of transfer function T_01 (between channels CS0 and CS1), and similarly also transfer functios T_02 and T_03,
    • estimations of impulse responses h_01, h_02 and h_03. Compare previously computed delay using CCF or CPSD and position of the maximum of computed impulse response.

  • Realize the modeling of room reverberations using estimated impulse responses by the filtering of the signal SA001S04.CS0.
  • Result:
    • Observe spectrograms of (real) signals from channels CS0, CS1, CS2 and CS3 (recorded real signals of other channels are SA001S04.CS1, SA001S04.CS2 and SA001S04.CS3) and compare them with spectrograms of modeled signals CS1-mod, CS2-mod and CS3-mod.
    • Real and modeled signals could be compared also by illustrative listening.